mRNA returns to cancer, a technology that took sixty years to get here

During the pandemic many people believed messenger RNA technology had just been invented to make COVID vaccines. In reality, the technique for producing it in vitro had been making its way for decades and had already been tested against cancer. Now a personalised therapy has made the leap to phase 3 in melanoma, while others are advancing in very different tumours. The news deserves enthusiasm, but not triumphalism. The full data still matter, and the pancreatic result remains a promise yet to be proven, where US politics adds a paradox that is hard to ignore.

On 19 August the American pharmaceutical company Moderna, which everyone will know as the maker of one of the vaccines we were given during the pandemic, announced that the large INTerpath-001 trial had met its objectives in patients with resected melanoma at high risk of relapse. What is truly new is that the treatment, intismeran autogene, is manufactured individually for each patient from samples of their own tumour and is given together with pembrolizumab, an immunotherapy from Merck that is already in use. It is the first time a personalised mRNA cancer therapy has delivered a positive result in phase 3. Those last two words, phase 3, are worth underlining, because until now the technology had accumulated promising signals but had not passed a test of this size.

A positive phase 3 changes the category of the evidence, but it does not yet turn the promise into a cure.

Nor do we yet know everything we would like to know. The announcement is preliminary; the full numerical results, the relapse curves, the confidence intervals, the subgroup analyses and, above all, the final effect on survival are still missing. The 49% figure that is often quoted comes from the earlier phase 2 trial and expresses a relative reduction in the risk of relapse or death; it does not mean the vaccine cures 49% of patients. Phase 3 has been positive, which is a milestone, but carrying over figures from another study would be confusing enthusiasm with precision.

The second piece of news comes from the cancer with the lowest five-year survival of all those treated: pancreatic cancer. In an initial study, sixteen patients who had undergone surgery received autogene cevumeran, another personalised mRNA vaccine, in this case from a different company, BioNTech, in addition to other treatments. During COVID, when people talked about the Pfizer vaccine, it was really BioNTech’s work.

Eight developed a strong immune response and, after several years, seven were still alive; among the eight without that response, two survived. The difference is striking, but the study did not compare vaccine with placebo, but responders with non-responders within the same group. With eight people on each side it is impossible to know whether the vaccine caused the advantage or whether those who responded also had a better prognosis for other reasons.

In pancreatic cancer we have a fascinating biological signal; we do not yet have a demonstration of efficacy.

The real answer will have to come from the randomised phase 2 trial already under way with some 260 patients. That caution does not diminish the interest of the finding; rather, it puts it where it belongs. Pancreatic tumours usually have fewer exploitable mutations than melanoma and raise a barrier around themselves that makes it hard for the body’s defences to get in and act. Achieving durable immune memory in that context is scientifically remarkable; proving that it prolongs life will be something else, and considerably harder.

To understand how we got here we have to go back to the 1960s, when mRNA was identified in the cell as the temporary copy that carries the instructions for making a protein from the DNA.

The therapeutic idea arose early. If that was the genomic mechanism for transmitting information for cell replication, why not write a message, introduce it into the cell and ask it to produce, for a while, whatever we needed? The concept was clean; the reality, not so much.

mRNA is not a specific vaccine; it is an information carrier capable of giving different instructions.

Researchers quickly discovered that mRNA degrades easily, entered cells poorly and could trigger a defensive reaction from the body’s own macrophages, which attack it and therefore halt the desired production.

During the 1980s and 1990s several teams learned to manufacture it and introduce it into cells; by 1995 a strategy against a tumour antigen was already being tested in animals. So the use of mRNA in oncology did not appear during COVID. It was the other way round: thanks to the previous 25 years of research, a vaccine could be developed in record time.

This is where Katalin Karikó comes in, and credit should be shared without creating a single heroine. Karikó did not discover mRNA or its possible applications. She was not the first person to introduce it into a cell, nor did she develop on her own the lipid systems that would later make it possible to transport it. Her contribution, together with Drew Weissman and other collaborators, solved one of the decisive obstacles: they showed that certain modifications of its components could reduce the unwanted inflammatory reaction and allow the message to produce more protein. The fundamental work was published in 2005 and was refined afterwards. This needs explaining, because when these advances were beginning, in the 1980s, I was at university studying these things. The side effects were so brutal that my professors used to say “they make the fillings jump out of your teeth”. In other words, until Karikó and Weissman modified it, the discovery was useless. That is why, although we write it the same way, the therapeutic kind is called modified mRNA.

Karikó and Weissman did not invent the platform on their own; they solved one of its fundamental bottlenecks.

That is what the 2023 Nobel Prize in Medicine recognised. Without the earlier science there would have been no platform and, without the later advances in purification, manufacturing and delivery, there would have been no effective vaccines either. Nanoparticles protected the message and helped get it into the cell; sequencing and computing made it possible to read a tumour and choose its targets.

In melanoma, the process begins with the laboratory analysis of the malignant cells of the removed tumour, comparing their genome with that of a healthy sample from the same patient. Comparing both sequences reveals mutations in the DNA of the malignant cells that belong only to the cancer. A computer system, using AI, selects those most likely to be recognised by the body’s defences, and an mRNA is manufactured that can include up to 34 signals. Once administered, the patient’s cells temporarily produce those signals and teach the immune system which malignant cells to look for. In this case, Merck’s drug, pembrolizumab, has the job of preventing the tumour cells from activating the brakes that switch off those new defences. Put simply, it stops them from disabling the new infantry. One part points to the target; the other helps attack it.

For the first time a medicine can be made not for a disease, but for the specific tumour of a specific person.

That personalisation is also the difficulty. It makes manufacturing more expensive, complicates logistics and does not stop the tumour from evolving. But you have just taken a problem that was, until now, unsolvable and turned it into a problem of logistics and money.

How many treatments of this kind, for other cancers, have already reached advanced stages of study and analysis? The answer changes depending on whether we count molecules, indications or trials. On a conservative basis there are today at least five mRNA cancer therapies in phase 2 and two in phase 3. That is, seven different types of cancer. This has never been seen before with a single technology.

Five different candidates are in phase 2 and two in phase 3; counting trials or indications produces much larger figures.

The difference is not a statistical detail. Moderna is studying several settings in lung, kidney and bladder cancer and metastatic melanoma. But some tumours have sub-indications, so we could properly say that Moderna has nine advanced programmes. BioNTech and Genentech are testing in pancreatic cancer, colorectal cancer, advanced melanoma and head and neck cancer associated with the human papillomavirus.

This landscape explains why the US political debate matters more than it seems. On 5 August 2025 the Department of Health announced the cancellation, reduction or restructuring of 22 federal investments in mRNA vaccines, worth close to 500 million dollars. The decision concerned infectious diseases, and the announcement itself made clear that other uses of the technology were not affected. It would therefore be false to claim that the Trump administration has cancelled mRNA cancer vaccines. It has not.

Trump has not cancelled mRNA cancer research; the risk, if there is one, is indirect.

The reasonable criticism is less immediate and perhaps more serious. Technologies do not respect budget compartments. A plant that learns to make mRNA for flu acquires capabilities useful for oncology; an improvement in stability, delivery or quality control serves more than one disease; scientists and technicians move between projects. Shrinking part of the market and of the funding can reduce scale, investment and learning across the whole ecosystem, even if none of the withdrawn items was earmarked for intismeran. That is an economic and technological possibility, not a proven causal link.

It is legitimate to debate doses, safety, duration, age groups, costs and priorities; it is not legitimate to turn the limitations of one application into a verdict on the whole platform.

Judging all of mRNA by a single application would be like judging all drugs by the result of one.

Karikó’s experience counsels caution in the face of final verdicts. A platform that seemed unviable ended up solving problems that no one had solved all at once.

What is decisive may not be the word vaccine, but the transformation of the medicine itself. The tumour is read and a sequence is made for that person. If it works, it will change regulation, trials and cost.

The medicine is no longer just a substance; it is starting to be personalised information.

None of this guarantees that mRNA will transform every cancer treatment. Melanoma is favourable ground for awakening the body’s defences, the full phase 3 data have yet to be published, and the pancreatic result rests for now on sixteen patients. Individualised vaccines are expensive, slow and complex; some of the chosen targets will not work and others will disappear as the tumour evolves. Relapse-free survival will also have to translate into a sufficiently large improvement in overall survival and quality of life.

But we are no longer looking at a laboratory curiosity. There are seven different treatments that have reached phase 2 or 3 under the strictest criteria, multiple tumours under study and a first positive phase 3. Sixty years after the messenger was discovered, science is starting to use it to write made-to-measure medicines. Some bets may fail; that will happen. What would be reckless is to confuse that uncertainty, which belongs to all serious research, with the certainty that we already know the limits of a technology that has just crossed its most important clinical frontier.

One would hope that the American mistake on funding would be picked up immediately by the EU. The general public has now been shown what the medicine of the future will be.

It is the duty of politicians, however ignorant they may be, and they are, to allocate adequate funds so that these advances keep moving forward, since those funds do not come out of their pockets but out of ours. But allow me to pour a little cold water. Seeing some people’s dedication to the public good, their tremendous ignorance of everything, their utter lack of management ability and the scant empathy of others, I fear the worst. And I am not only referring to the Americans, but to others closer to home.