How vaccines work

How do vaccines work?

A vaccine protects you from a disease — yet it does not fight the disease for you the way a medicine might. So what does it actually leave behind in your body? If it is gone within days, how can it guard you years later?

Consider what happens the first time you catch an illness: you are sick for a while, then recover, and often you never catch that exact illness again. Your body seems to learn. So what if a vaccine let the body do the learning without the being-sick part? What would it have to show the body to teach it?

Start with why the first encounter goes badly. Your immune system carries no defence against measles specifically; it carries an enormous library of cells, each recognising one essentially arbitrary molecular shape. When a pathogen arrives, the few cells whose shape happens to fit must be found among millions and then multiplied, and that search-and-multiply takes a week or two — the illness being what happens meanwhile, while the pathogen has the field to itself. Ask what would change if the search had already been done and the winners kept on file, and you have the whole idea.

What gets kept? Two lineages doing different work. B cells make antibodies — proteins that latch onto the pathogen's surface to mark it, or block it from entering your cells at all. T cells split further: some direct and license the rest of the response, and some patrol for your own already-infected cells and kill them, which is the only way to reach a virus that has got inside. Antibodies handle what is loose in the blood; killer cells handle what is hiding. A vaccine that trains the first arm well and the second less so may prevent serious illness while doing much less about catching and passing the thing on.

And what does the memory physically consist of? Not a record of the event but a changed population. The cells that responded were selected and multiplied, so there are far more of them than before; they have been through a refinement that leaves their antibodies binding the target much more tightly; they act faster on a second sighting; and some settle into your bone marrow as long-lived factories, quietly secreting antibody for years with no pathogen present. Memory is not stored somewhere — memory is that surviving population.

Which raises the follow-up: if memory is a population of cells, protection lasts as long as that population does and as long as the target stays recognisable. Two doses of measles vaccine are about 97 percent effective and taken to protect for life; tetanus needs a booster every ten years or so; influenza is reformulated annually. Only the middle case is really about memory fading. Measles barely changes, so a memory learned once still fits; influenza changes its surface constantly, so the memory is fine and the face is different. Those are not the same failure, and lumping them together is the source of much bad reasoning about vaccines.

So the vaccine itself does no fighting and soon disappears. What remains is the memory — immune cells trained on the face of the threat. When the real pathogen arrives, the response is fast and prepared instead of slow and improvised.

Here is the misreading that matters most, because it turns a success into a suspicion. A sore arm, a headache, a night of feeling flattened — people take these as the vaccine harming them. They are not. They are your own inflammatory response: the local signalling that summons cells to the injection site, and the fever-and-fatigue chemistry accompanying any immune activation. The discomfort is the learning, which is why it arrives a day later rather than immediately, and why it is often stronger after a second dose. It follows, awkwardly for intuition, that feeling nothing is not evidence the vaccine failed; how rough you felt tracks how well you are protected only weakly.

A second thing the poster hides is that the immune system largely ignores a face presented on its own. Show it a purified protein with no sign of danger attached and it may conclude nothing is wrong and decline to commit. So most non-live vaccines include an adjuvant — classically an aluminium salt, used since the 1930s — to supply the alarm that makes the response worth mounting. Honest footnote: after roughly ninety years of use, how aluminium adjuvants actually work is still not fully settled, which is a fair illustration of how much of immunology is empirical.

Two smaller corrections. A live weakened vaccine and a killed or subunit one are not the same instrument: the live kind more closely resembles a real infection and tends to produce broader, longer-lasting memory, which is part of why some need no boosters at all. And "immunity" oversells it — most vaccines shift the odds and the severity rather than sealing a door. One clause on the collective side, since it deserves its own treatment: because a vaccinated person is a worse host to pass the pathogen on, enough of them slows transmission for the people who cannot be vaccinated at all.

• Why some vaccines need boosters while others last a lifetime.
• How "herd" protection shields people who cannot be vaccinated.
• Why a vaccine can prevent serious illness while doing much less to prevent transmission.

Every term the collection defines is gathered in the glossary.