Dr. Philip McMillan,  John McMillan

The switch fixes one of the strangest problems with mRNA vaccines and leaves a bigger one untouched.

In late April 2026, Sanofi announced that it had beaten Moderna and Pfizer in a Swedish national tender. From the 2027/2028 season, every COVID-19 vaccine in Sweden’s national programme will be Nuvaxovid, a protein-based shot originally developed by Novavax. No mRNA vaccine made the cut.

The decision is striking for another reason. Sweden ran this tender on its own, stepping outside the European purchasing agreements that have supplied most of the continent since the pandemic. Countries do not usually walk away from a shared contract and switch technologies at the same time.

So why now? Price may be part of it, but practical reasons are rarely the whole story. Could this be the first crack in the dam, a quiet signal that the mRNA approach has not worked as promised? And if so, does the replacement actually solve the problem?

What Sweden Bought

The contract, run by the purchasing body Adda on behalf of all 21 Swedish regions, covers everyone aged 12 and over from June 2027. Sanofi’s announcement leaned on proven technology, the same approach used for years in hepatitis B and HPV vaccines, plus a prefilled syringe that keeps in an ordinary refrigerator.

In June 2026, Sweden’s Public Health Agency signalled a further shift. Since November 2022 it had advised against Nuvaxovid for people under 30, after Australian data hinted at higher rates of myocarditis, an inflammation of the heart muscle, in younger recipients. That suspicion, an agency physician told the Swedish medical press, no longer applies in the same way, and the age limit is set to be relaxed.

Swedish coverage spelled out the difference between the platforms in a single line. With mRNA vaccines, the body is instructed to make viral proteins itself. A protein-based vaccine contains a finished part of the virus for the immune system to practise on. That sentence amounts to an official acknowledgment that with mRNA, nobody controls how much spike protein a given body ends up making.

A Dose Nobody Could Measure

Consider how an mRNA vaccine works. Strands of messenger RNA, wrapped in tiny fat bubbles called lipid nanoparticles, are injected into the arm. Nearby cells take them up and feed the instructions into their ribosomes, the protein-making machinery inside every cell. Out comes spike protein, displayed on the cell surface to provoke an immune response.

The catch is that nobody knows how much spike any one person makes. A strand of mRNA might be read a handful of times in one body and thousands of times in another. And the material does not always stay in the arm: pieces of vaccine mRNA have been found in blood weeks after injection. The dose of spike is presumed, never measured.

A protein vaccine turns that around. The spike is made in advance and paired with an adjuvant called Matrix-M, an ingredient that rouses the immune system to respond. What goes into the arm is exactly what the immune system sees, and then it is broken down.

That difference may explain one of the most baffling features of the rollout. Some people had severe reactions to mRNA shots, while others barely produced antibodies. If spike output varies wildly between individuals, both extremes make sense. A known dose takes that variable off the table.

Platform or Protein?

Nuvaxovid is no newcomer. In a placebo-controlled trial in the United Kingdom, published in late 2022, two doses showed 89.7 percent efficacy against COVID-19, with markedly fewer cases of infection and severe disease than placebo. Protection against severe illness was never the main worry. The harder question is what happens when vaccinated people are infected again, and that has never been properly answered.

A second puzzle points toward the platform. In 2024, a study in Nature Medicine examined bone marrow, home to the long-lived plasma cells that churn out antibodies for years. Cells against influenza and tetanus were easy to find there. Cells against SARS-CoV-2, after mRNA vaccination, were largely missing, which helps explain why antibody levels fell away within months. A quirk of mRNA delivery, or something about spike itself? Sweden’s switch could become the natural experiment that answers it.

Locked on the Original

Even a perfect platform runs into a second obstacle, known as immune imprinting. The first vaccines were built on spike from the original Wuhan strain. When Omicron arrived with a reshaped spike, manufacturers updated the formula and expected the immune system to update with it.

It did not, at least not fully. Shown the new spike, the immune system keeps reaching for its first memory and pumps out antibodies tuned to Wuhan. Studies of updated boosters have found the response still dominated by recall of those original memory cells. Changing the protein in the vial does not reset the memory in the body.

That is why spike-based vaccination looks increasingly ill-suited to stopping the virus from circulating. Nasal sprays that train mucosal immunity, the defences lining the nose and throat where infection begins, were an obvious alternative. So were vaccines built around other viral proteins. Perhaps those proved harder to make. Either way, the field stuck with spike.

The Factory, Not the Parcel

Natural infection teaches a different lesson. A coronavirus carries several proteins besides spike, among them the membrane, envelope and nucleocapsid proteins. When the virus takes over a cell and starts producing all of them, the cell is obliged to display fragments of every protein it makes on its surface. Think of it as a built-in audit system: nothing gets made without being shown to the outside world.

Killer T cells read those displays. After natural infection, only around a quarter of the T-cell response is aimed at spike. The rest spreads across the virus’s other proteins, and any cell showing any of them becomes a target.

“Antibodies will target the protein. The T cell targets the factory, which is the cell,” said Dr. Philip McMillan, a British researcher and clinician.

A single infected cell can release thousands of new viral particles. Antibodies chase those particles one at a time, while T cells shut down production at the source. Broad T-cell immunity, working where the virus first lands, goes a long way toward explaining why natural immunity is so effective. Training the immune system on spike alone leaves most of that defence on the bench.

The Question Sweden Didn’t Ask

By choosing a known dose of a familiar kind of vaccine, Sweden has fixed a real flaw. It has not changed the target. Nuvaxovid still teaches the immune system to recognise spike, the protein most closely tied to reported vaccine side effects and to many patterns seen in long COVID.

Remember, too, what killed most people with severe COVID-19. Sometimes a viral pneumonia did the damage. In the vast majority of cases, the culprit was the cytokine storm, the immune system turning on the body’s own tissues. That alone argues for caution in how vaccines provoke it.

Vaccinating high-risk people who had not yet met the virus was a reasonable call. Extending it to entire populations, dose after dose, is another matter, with consequences still playing out.

“Now, if you acknowledge the spike protein is the driver for most of the pathology, why would you continue to expose the immune system to it?” said Dr. McMillan.

The usual answer is that the virus, not the spike, causes the harm. But if that were true, why would injecting spike alone produce some of the same problems? Something does not add up. Sweden has shown it is willing to change platforms. The braver step would be to go back to basics and ask whether spike is the wrong protein for any platform at all.