Japan Advanced Institute of Science and Technology (JAIST) researchers report complete tumor elimination in mouse colorectal cancer using IV Ewingella americana, framing the dual mechanism as a candidate platform for bacterial cancer therapy.
An intravenous dose of a bacterium isolated from the intestines of a Japanese tree frog eliminated every colorectal tumor in a mouse model. The result gives bacterial cancer therapy a defined mechanism to point to after decades of partial-regression studies.
The bacterium, Ewingella americana, lives in the gut of Dryophytes japonicus. Researchers at the Japan Advanced Institute of Science and Technology (JAIST) injected a single dose into mice bearing colorectal tumors and recorded a 100% complete response rate, according to findings published in the peer-reviewed journal Gut Microbes and detailed in a December 2025 institutional press release. The ScienceDaily writeup and a SciTechDaily re-report carried the result into general-audience news in early July.
The mechanism is what makes the result matter beyond a curiosity. Ewingella americana is a facultative anaerobe, a class of bacteria that can survive and multiply in low-oxygen environments. Solid tumors develop oxygen-starved cores that most chemotherapy drugs struggle to reach. In the mouse model, the bacterium colonized those hypoxic cores and acted on the cancer cells directly, while also appearing to activate the host immune system against the tumor. The authors describe this as a dual attack: bacterial cytotoxicity plus immunomodulation.
The mouse experiments also benchmarked the approach against two existing therapies. The bacterium outperformed both an anti-PD-L1 immune checkpoint inhibitor antibody and the chemotherapy liposomal doxorubicin in the same colorectal cancer model. That is a mouse-versus-mouse comparison, not a replacement claim for any approved human treatment, but it does establish that the effect is not trivially explained by the immune system doing the work on its own.
Earlier studies have tested fecal transplants or dietary interventions aimed at reshaping the microbiome. The JAIST team isolated individual bacterial strains and delivered them IV to act on tumors directly, which sidesteps the variability of whole-microbiome therapies and gives researchers a single agent to characterize.
The New York surgeon William Coley injected bacterial toxins into inoperable tumors in the 1890s and recorded some regressions, but the approach never produced a standard treatment, in part because the active agent and the mechanism stayed unclear. The modern revival has leaned on engineered Salmonella, Listeria, and other strains tuned to deliver payloads or trigger immune responses. The JAIST result is unusual in that the strain is unengineered, naturally occurring, and the proposed mechanism runs through a single anatomical feature: the hypoxic tumor core.
This is not a treatment for human cancer. The data is preclinical, in a single mouse colorectal cancer model. Ewingella americana has prior history as an opportunistic human pathogen, particularly in immunocompromised hosts, and the JAIST team has not yet published mammalian safety data showing that an IV dose of a live frog-derived bacterium is tolerated by larger animals, let alone by people. Tumor type generalization, dose translation, and pharmacokinetics are also open. The authors frame the work as a candidate platform that could extend to other solid tumors, but that is forward-looking, not demonstrated in this paper.
The team says the next moves are independent replication in other labs, mammalian toxicology in larger animals, and the early safety work that precedes any first-in-human study. The path is long, but the mechanistic map for it is now in print.