Before Ewingella americana was a promising cancer microbe from a Japanese tree frog, it was a name in hospital microbiology reports. The species was first described in 1983 by Grimont and colleagues, who characterized it from ten clinical strains isolated in the United States. It was first found in sick people. The frog came 42 years later.
That history is largely missing from coverage of the striking preclinical result from the Japan Advanced Institute of Science and Technology, where a single intravenous dose of E. americana eliminated colorectal tumors in every treated mouse, outperforming both liposomal doxorubicin and an anti-PD-L1 checkpoint inhibitor. MedicalDaily reported those findings last month. The clinical file on the organism is what will determine whether any of it reaches patients.
Nineteen Documented Patients, Nine with Bacteria in the Blood
A narrative review published in the journal Antibiotics by researchers at the University of Crete assembled every published human case they could find. Sixteen studies described 19 patients. The median age was 55.
Bloodstream infection was the most common presentation, in nine patients. Three had peritoneal dialysis-associated peritonitis. Three had lower respiratory tract infections. Two had conjunctivitis. Single cases involved the urinary tract, a bone and joint, and the central nervous system. Fever appeared in 57.9% of patients.
Among the nine bacteremia patients, 77.8% acquired the infection in a hospital, and two-thirds had undergone surgery within the previous three months, most often cardiac surgery. One patient, a 74-year-old woman, died of the infection.
The cases skew toward exactly the populations that bacterial cancer therapy would be aimed at. Three of the 19 were immunosuppressed, including recipients of a kidney transplant and an allogeneic bone marrow transplant. A 2025 case report from Türkiye described a man with testicular choriocarcinoma who became septic during a red blood cell transfusion on the second day of chemotherapy. Blood cultures grew E. americana. He recovered on ceftriaxone.
The Organism Also Lives in Hospitals, and in Salad
E. americana has been recovered from plants, cultivated mushrooms, mollusks, vacuum-packed meat, and salad vegetables. It has also shown a talent for colonizing clinical environments. In 1987, investigators at a U.S. hospital worked up 20 patients apparently bacteremic with the species and concluded it was a pseudo-epidemic traced to contaminated collection tubes and non-sterile blood collection technique. No one was truly infected, but the organism had clearly established itself in the building.
Comparative genomics of sequenced strains, published in 2020 in the journal Pathogens, identified 27 distinct antibiotic resistance genes across sequenced E. americana genomes, spanning four resistance pathways, with multi-efflux pump families being the most heavily represented. The same analysis identified genes for virulence effectors, secretion systems, toxins, and pigments associated with evading host defenses. The authors were explicit that pathogenicity studies have not been done, and that the genetic findings imply an opportunistic pathogen rather than prove one.
The Strain Question Nobody Has Answered Yet
None of this invalidates the JAIST work, and it is important not to overstate the conflict. Species-level identity is not strain-level identity. Isolates from the same species can differ substantially in virulence gene content, resistance profiles, and host behavior, and the frog-derived isolate has not been shown to carry the features documented in the clinical strains.
The JAIST safety data in mice were reassuring on their own terms. The bacteria cleared from the bloodstream within 24 hours, did not colonize the liver, lungs, or kidneys, and produced only mild transient inflammation across 60 days of observation.
But those mice were immunocompetent. The published human infections cluster in patients who were not: people recovering from cardiac surgery, on dialysis, on immunosuppressants, or mid-chemotherapy. Overall mortality across the 19 documented cases was low, at one death, and susceptibility to aminoglycosides, cephalosporins, and trimethoprim-sulfamethoxazole was high, which is genuinely favorable if an infection ever needed treating.
Why Bacterial Cancer Therapy Keeps Running Into This
The field has always lived on this edge. William Coley induced tumor regression with bacterial preparations in the 1890s and lost the approach partly to infection risk. BCG, a live attenuated mycobacterium, remains standard therapy for non-muscle-invasive bladder cancer more than three decades after its 1990 FDA approval, and it can cause disseminated infection.
Any candidate potent enough to colonize a hypoxic tumor and provoke a systemic immune response is, by construction, a candidate to do something unwanted in a patient with a damaged immune system. That is the trade being negotiated, not a flaw someone overlooked.
What a serious development program would need before human testing is strain-level genomic characterization against the clinical isolates, susceptibility testing on the therapeutic strain, work in larger animals, and a defined plan for what happens if a treated patient becomes bacteremic. None of that has been announced. No clinical trial exists.
Patients with colorectal cancer should not read this as a treatment on the horizon and should not alter any part of their care based on animal research. Anyone eligible for colorectal cancer screening, which the CDC recommends beginning at age 45 for average-risk adults, should talk to a clinician about it.
Key Questions Answered
What is Ewingella americana?
A gram-negative bacterium in the family Yersiniaceae, first described in 1983 from ten clinical strains in the United States, and the only species within its genus.
Has it made people sick?
Yes. A 2024 review found 19 published cases, most commonly bloodstream infections, mostly hospital-acquired.
Does that mean the cancer research is dangerous?
Not necessarily. The frog-derived strain has not been shown to carry the virulence or resistance features of clinical isolates, and strains within a species can differ considerably.
Who is most vulnerable to infection with it?
Documented cases concentrated in people after recent surgery, on peritoneal dialysis, on immunosuppressive drugs, or receiving chemotherapy.
Can it be treated?
In the published cases, yes. Resistance to aminoglycosides, cephalosporins, and trimethoprim-sulfamethoxazole was minimal, and one of 19 patients died.
Is any of this available to patients?
No. The tumor findings are from mice, and no human trial has been announced.
