The immunotherapy trials are multiplying. Cancer first, then autoimmune disorders, then allergies, then brain disease. The Guardian ran an explainer on immunotherapy: what it is, how it works, which conditions it might treat. The article opens with the phrase “the body’s defences.” I read that and stopped. Not because it’s wrong. Because it reveals what gets designed into the treatment before the first trial begins.
A defence system has enemies. It has borders. It has mechanisms for identifying what belongs and what does not. This is not metaphor. This is the conceptual architecture that determines which bodies get legible as treatable.
My immune system attacks my joints. Has done since I was sixteen. The rheumatologist explained it clearly: my body mistakes itself for an invader. The treatment suppresses the mistake. What nobody said — what the treatment design makes invisible — is that “mistake” and “invader” are spatial concepts imported from military thinking. The immune system is not neutral biology. It is a system we named after diplomatic immunity, after the concept of a sovereign body that can expel what does not belong.
Immunotherapy works by teaching the immune system to recognise cancer cells as foreign. The language is precise: recognition, targeting, elimination. What this means in practice is that the treatment succeeds when your body can be made legible as a space with clear borders. When the system can identify what should be inside and what should not.
I have sat in infusion clinics where the chairs face each other in rows. You watch the person across from you receive the same drug through the same port for a different autoimmune condition. The nurse checks your blood work first — they are looking for signs that your immune system has been suppressed past the point where you can fight off what actually threatens you. This is the trade. You take a drug that stops your body attacking itself. The drug also stops your body attacking anything else.
The trial design does not include this trade in its concept of efficacy. Efficacy means: did the drug do what we designed it to do? Did it suppress the immune response we targeted? The fact that it also suppresses immune responses you need to stay alive is filed under “side effects.” Side effects are not part of the system. They are errors in the patient.
There is a researcher named Polly Matzinger, an immunologist at the National Institutes of Health, who proposed in the 1990s that the immune system does not respond to “foreign” substances. It responds to danger signals. Tissue damage. Stress. Cells dying in ways that indicate threat. Her model — called Danger Theory — never replaced the self/non-self framework that dominates immunology textbooks. The self/non-self framework is cited frequently. It has not changed how treatments get designed.
I keep thinking about why. The danger model would require designing treatments that account for context. What is dangerous in one body or one moment may not be dangerous in another. The self/non-self model allows universal protocols. One drug, one dose, one mechanism. It is easier to test. Easier to approve. Easier to manufacture at scale.
Easier, but not for the body receiving it.
The immunotherapy explainer lists conditions these treatments might address: cancer, autoimmune disease, allergies, infections, brain disorders. Five categories that have nothing in common except this: they all involve immune response. The article presents this as progress — one mechanism, many applications. What it does not say is that grouping these conditions together redefines what the problem is. The problem is not cancer. The problem is not rheumatoid arthritis. The problem is immune response itself.
This is not a medical argument. This is an architectural one. If you design a treatment around the concept of borders, you will produce therapies that strengthen borders. The drug I take works. It stops my joints from destroying themselves. It also means I get every cold, every flu, every infection that moves through a room. I spend two weeks recovering from things that used to take two days. This is not a side effect. This is the treatment working exactly as designed.
The design assumes the borders were the problem. The design does not ask whether borders are the right way to think about a body at all.
There is a building in Rotterdam — the Sonneveld House, designed in 1933 — where the architect installed a system of sliding walls. Rooms reconfigure depending on use. The dining room becomes part of the living room. The study closes off or opens up. The walls are not fixed borders. They are thresholds that move.
I thought of that house the first time someone explained to me that autoimmune disease means your body has lost the ability to distinguish self from non-self. As if the self were a fixed territory. As if the immune system’s job were to patrol a border. The Sonneveld House suggests a different model: spaces that respond to need, boundaries that shift, a structure designed for change rather than defence.
Immunotherapy research is not designed this way. It is designed to restore the border. To teach the immune system to target correctly. “Correctly” means: according to a model of the body as sovereign territory.
I have read trial results where efficacy is measured by reduction in immune response. The drug worked if your body stopped attacking itself. What the trial does not measure: whether you can still recognise danger. Whether you can still fight infection. Whether the borders they restored are borders you can live with.
The Guardian article ends with optimism. A wave of trials. New applications. Hope for conditions that have resisted treatment. I do not doubt the hope is real. I doubt the hope is reading the room.
What is being designed right now is not a treatment for your body. It is a treatment for the concept of a body as a system with defensible borders. The immunotherapy model assumes that if the borders can be clarified, the problem resolves. What this misses is that some bodies do not operate on borders. Some immune systems respond to danger, not foreignness. Some conditions cannot be mapped onto a model of self and invader without losing what makes them legible in the first place.
The trials are multiplying. The mechanism is elegant. The results, for some people, are transformative. None of that changes what the treatment is designed to see. And what it is designed to see determines who it can treat.
The Sonneveld House still stands.