A cancer vaccine with an audience of one. They read your tumor's mutations and build the drug around them.
In a lab outside Philadelphia, a sequence of a patient's tumor is being turned into a product that has never existed and will never exist again. Not a pill pulled off a shelf. Not a one-size dose. A ring of DNA written specifically for the mutations inside one person's cancer, and useless to anyone else on Earth. This is what Geneos Therapeutics does on an ordinary Tuesday.
The company is small - around 13 people - and it works on the least small problem there is. Cancer hides by looking like you. Geneos's whole bet is that the places where it stops looking like you, the mutated proteins called neoantigens, are exactly where the immune system can be taught to attack. Find those typos. Print them into a vaccine. Hand the immune system a wanted poster.
It sounds like marketing until you read the data. In 2024, Geneos put its results in Nature Medicine - the kind of journal that does not hand out pages for press releases. The trial was small. The claim was not.
The last decade of oncology has a hero: the checkpoint inhibitor. Drugs like pembrolizumab take the brakes off the immune system and let it see cancer again. For some patients, it is close to miraculous. For most, it does nothing - because taking the brakes off only helps if there's a car pointed in the right direction.
That is the gap Geneos exists to fill. A checkpoint inhibitor releases the immune system; it doesn't tell it what to chase. Without specific T cells trained on the tumor's own mutations, the immune system is unleashed and aimless. Liver cancer - advanced hepatocellular carcinoma, the company's lead target - is one of the toughest rooms in the building, with stubbornly low response rates and few second-line options.
Translation for the rest of us: the immune system is a guard dog. Checkpoint drugs unclip the leash. Geneos hands it the scent.
Niranjan Sardesai had a comfortable view of the problem. As Chief Operating Officer of Inovio Pharmaceuticals, he had helped build a company with cancer and infectious-disease immunotherapies in late-stage trials. He had a PhD in chemistry from Caltech and an MBA from Wharton - a combination that usually leads to a board seat, not a startup with a dozen employees.
In 2016 he left to start Geneos anyway, on a contrarian premise. While the field sprinted toward mRNA, Sardesai bet on DNA plasmids - older, less fashionable, and, he argued, better suited to the job. A DNA plasmid can carry a lot of cargo. Geneos designed its vaccines to encode up to 40 of a patient's neoantigens in a single plasmid, and up to 80 by combining two. mRNA vaccines, the competition's weapon of choice, tend to carry fewer.
The wager was simple and uncomfortable: that personalization at this depth - dozens of targets, bespoke per patient, manufactured on demand - was worth the operational headache it created. Every patient is a new product. Every product is a small factory run. Most of biotech is built to avoid exactly this. Geneos was built to do it.
The platform is called GT-EPIC. Strip away the trademark and here's the loop: sequence a patient's tumor, identify the neoantigens that mark it as foreign, design a DNA plasmid that encodes them, and deliver it into the skin alongside a cytokine adjuvant - DNA-encoded IL-12 - that turns up the immune volume. The delivery itself uses electroporation, a quick electrical pulse that helps the cells take up the DNA.
Reads a tumor's mutations and designs a custom DNA plasmid encoding up to 40 patient-specific neoantigens - 80 with two plasmids.
The personalized therapeutic cancer vaccine itself, co-delivered with a DNA-encoded IL-12 cytokine adjuvant.
Injected into the skin of the arm and driven home with electroporation - the worst common side effect was a sore injection site.
Paired with an anti-PD-1 checkpoint inhibitor so the immune system is both released and aimed at the right targets.
The result is meant to be tumor-infiltrating lymphocytes - T cells that don't just circulate but actually traffic into the tumor, programmed by the identity of the neoantigens Geneos chose. The vaccine doesn't kill the cancer. It builds the thing that does.
Niranjan Sardesai leaves the COO seat at Inovio to build a company around personalized DNA cancer vaccines.
Led by KIP Global Bio Fund, with Sante Ventures and Inovio. Clinical updates on the personalized cancer vaccine program follow.
Geneos reports a second complete response in its ongoing trial in second-line advanced liver cancer.
An additional $5M tranche, including Shanghai Healthcare Capital, brings the round to roughly $10M.
Positive Phase 1/2 GT-30 data published - billed as the first definitive demonstration of a personalized vaccine enhancing anti-PD-1 response.
The GT-30 trial was a single-arm, open-label, multi-center Phase 1/2 study: 36 patients with advanced hepatocellular carcinoma who had already failed a standard tyrosine-kinase inhibitor. They received GNOS-PV02 plus DNA-encoded IL-12 plus pembrolizumab. The headline immunology result is the one that's easy to remember: in 14 of 14 evaluated patients - 100% - the vaccine induced new T cell responses against the neoantigens it encoded. Those T cells expanded in the blood and showed up inside the tumors.
The safety profile was, by oncology standards, gentle - the most common treatment-related side effect was a reaction at the injection site, in roughly 42% of patients. The efficiency signal came from circulating tumor DNA: at least a 50% drop in about 41% of evaluable patients, and every one of those responders correlated with ongoing survival.
For scale: roughly thirteen people, one journal that doesn't print favors, and a hundred-percent immunology hit rate in the patients they could measure.
Geneos states its mission without flourish: bringing personalized immunotherapies for cancer to the world, one patient at a time. The phrase is doing more work than it looks. "One patient at a time" is not a sentiment - it's a manufacturing constraint, a regulatory puzzle, and a logistics challenge wrapped in four words. A drug that must be rebuilt for every person can't be made the way drugs are usually made.
That is the company's bet on the future and its biggest risk at once. If personalization at this depth works - dozens of targets, per patient, fast - it points at a version of oncology where the question stops being "which approved drug fits this cancer?" and becomes "what does this specific tumor look like, and what should we build against it?" The harder question, the one investors keep asking, is whether you can do that at scale and at a price a health system will pay.
Planning has been underway for a potentially registrational trial in advanced liver cancer - the step that turns a striking Phase 1/2 result into something a regulator can approve. That's the next door. Whether Geneos walks through it is the open question.
Return to the scene. A sequence goes in. A ring of DNA comes out, written for a cancer that exists in exactly one body. A few years ago, that sentence was a hypothesis with a slide deck. After Nature Medicine, it's a result with a sample size - small, early, and real.
The skeptic's case is intact and worth respecting: 36 patients is not 3,600, a single arm is not a randomized trial, and "potentially registrational" is a phrase that has outlived many companies. Geneos has not cured liver cancer. What it has done is harder to dismiss - it has shown, in people, that a vaccine built from a patient's own mutations can teach the immune system something new and send it into the tumor.
The promise of personalized medicine has been "any day now" for twenty years. Geneos made a version of it that ships - one plasmid, one patient, one sore arm at a time. The vaccine in that lab isn't a metaphor for the future. It's an actual object, being made for an actual person, today. That's the whole point, and it's enough.