Engineered virus eradicates metastatic ovarian cancer in mouse models

An engineered virus wiped out metastatic ovarian cancer in mouse models, a striking preclinical result that puts protein design squarely in the middle of one of oncology’s hardest fights. The Stanford-led team built ErbB oncogene-selective virus, or ErbB-OSV, to replicate only in cells with abnormal ErbB signaling, a family that includes EGFR and HER2, then used that switch to destroy tumors while sparing healthier tissue.

The work, published June 12, 2026 in Nature Biomedical Engineering, was led by Xinzhi Zou, Elizabeth Palafox, Cynthia Zhao, Kevin T. Beier, Chil-Yong Kang and Michael Z. Lin. In plain terms, the researchers engineered proteins that act like molecular gatekeepers: the virus can copy itself only when it enters a cancer cell running on hyperactive ErbB signals. That matters because the paper says metastatic carcinomas such as ovarian cancer remain difficult to treat, in part because targeted drugs can damage normal tissue and immunotherapy still fails in too many patients.

ErbB-OSV posted better safety than a benchmark oncolytic virus from the same family, and it performed better against ErbB2, or HER2-positive ovarian cancer xenografts. In an advanced ovarian cancer syngeneic model, the strongest result came when the virus was paired with chemotherapy and repeated dosing was made possible by B cell depletion: that combination produced a 180% larger survival benefit than chemotherapy alone. The single-agent virus also cured most early cases in the mouse model, a result that will draw attention far beyond ovarian cancer because the strategy is built around signaling state, not a single tumor type.

That said, the road from mouse eradication to human treatment is still long. Safety will have to be proven in people, not just in preclinical models. Manufacturing will need to be consistent enough for a biologic that relies on engineered proteins and a viral backbone. Tumor targeting will need to hold up in the far messier biology of human metastatic disease, where cancers are heterogeneous and immune defenses can shut down repeat dosing.

Stanford researchers have chased metastasis with protein engineering before. In 2014, Jennifer Cochran and Amato Giaccia reported a decoy Axl protein that blocked Axl-Gas6 signaling and cut metastatic nodules in mice by 78% in breast cancer and 90% in ovarian cancer. That earlier technology was licensed to Ruga Corp. in Palo Alto, and it remains a reminder of the central question here: mouse victories can be real, but only a few survive the jump to patients. ErbB-OSV is an unusually elegant bet on that leap, not yet the leap itself.