Most medicines are manufactured by the million and pulled off a shelf. NeoVax is manufactured by the one. Each dose set is designed from the mutations in a single patient’s tumour, produced for that patient alone, and, in a phase 1 trial now published in Nature Medicine, combined with the checkpoint inhibitor nivolumab for people with platinum-sensitive recurrent ovarian cancer. The study asked whether that bespoke production line could work in practice.
That framing matters, because it defines what success means here. Success was a vaccine that could be made and delivered, a safety profile acceptable enough to continue research, and signs that the immune system responded. Success was not, and could not be in a trial of this size and design, a proven improvement in survival. Readers of our coverage of Starlink access decisions and their real-world effects know we separate the mechanism from the outcome; the same discipline applies to medical technology.
A production line with a batch size of one
The process begins with tumour tissue. Sequencing identifies mutations, software helps predict which of the resulting neoantigens, abnormal protein fragments unique to the cancer, are most likely to be shown to the immune system, and a vaccine is manufactured around those chosen targets. Every stage is individual: the input material, the design, the product and the recipient all belong to one person. Facilities and methods profiled by Dana-Farber Cancer Institute show how specialised this kind of personalised vaccine work remains.
That individuality is the approach’s promise and its problem. A vaccine matched to a patient’s own tumour mutations could, in theory, focus the immune response precisely where it is needed. But a batch size of one means no economies of scale, a manufacturing timeline that runs while the patient’s disease may be changing, and quality control that must be repeated for every single product. A feasibility trial is therefore partly an engineering audit: does the line run, end to end, for real patients?
Adding a checkpoint inhibitor to the system
Nivolumab blocks PD-1, a checkpoint that can restrain T cells. In this combination, the vaccine’s role is to teach or expand T cells that recognise the tumour’s neoantigens, while nivolumab’s role is to reduce the chance those cells are promptly restrained. It is a two-part system: one component supplies the target information, the other adjusts the operating conditions under which immune cells work.
In the trial, that system was delivered to patients whose ovarian cancer had recurred after a period of response to platinum chemotherapy. The researchers reported that manufacture and administration were achievable, that safety was acceptable in this small cohort, and that T-cell responses were observed. Those are system-level results: the machinery ran, patients tolerated it, and the intended biological signal appeared in measurements.
What the study cannot tell us yet
A phase 1 trial without a comparison group cannot separate the vaccine’s contribution from nivolumab’s, from the natural variation of the disease, or from the characteristics of patients well enough to enrol and wait for a personalised product to be made. It also cannot quantify rarer side effects. Anyone reading the Nature Medicine report as evidence that this combination helps patients live longer would be reading past its design, and past its authors’ intent.
That is not a dismissal. In technology development, demonstrating that a complex system can be built and operated safely is the gate through which everything else must pass. Many promising cancer ideas fail at precisely this gate, because they cannot be manufactured consistently or cannot be given without unacceptable toxicity. Clearing it earns a more rigorous test, nothing more and nothing less.
The next engineering and clinical problems
The next stage must solve two problems in parallel. Clinically, a larger, controlled trial needs to test whether adding NeoVax to checkpoint inhibition changes outcomes that matter to patients, such as tumour control and survival. Operationally, the production process needs to become faster, cheaper and more scalable, or even a positive clinical result would be difficult to deliver beyond a handful of specialist centres.
Both problems are now well defined, which is itself a form of progress. A decade ago, a vaccine designed from an individual tumour’s mutations and manufactured for one patient was largely a concept. This study shows the concept operating in a real clinic, in a hard-to-treat cancer, with acceptable safety and measurable immune responses. Whether it also works as a treatment is the question the next generation of trials exists to answer.
