Cancer Immunotherapy: How the Body Learned to Fight Back

For most of the twentieth century, oncologists fought cancer the way armies fought World War I: with poisons, radiation, and blunt instruments, hoping to destroy the enemy before destroying the patient. Then, in a series of slow discoveries spanning forty years, a different idea took hold — one that did not ask how to kill cancer but asked instead: why isn't the body already killing it?

The answer, when it came, was startling. The immune system can kill cancer. In most people, it does so constantly, destroying aberrant cells before they take root. The problem is that cancer, as Siddhartha Mukherjee puts it, is "an astonishing perversion of the normal cell" — one so successful at exploiting the body's own machinery that it has learned, among other tricks, to impersonate the signals of restraint that the immune system uses to prevent it from attacking healthy tissue. Cancer doesn't just evade the immune system. It subverts it.

The Emperor's Old Trick

Mukherjee — oncologist, cell biologist, Columbia professor, and author of The Emperor of All Maladies — has a DPhil in immunology from Oxford. He spent years studying how the immune system tells friend from foe, and why that discrimination sometimes fails. That background makes him unusually well-placed to explain why immunotherapy is not merely a new class of drugs but a conceptual revolution.

The core problem, he explains, is the checkpoint. The immune system runs on a series of molecular brakes — proteins that tell activated immune cells to stand down, to prevent them from turning on the body's own tissues. These are not design flaws; they are essential safety mechanisms. Without them, autoimmune disease is the result. The brakes have names: CTLA-4, PD-1, PD-L1. They sit on the surface of T cells and on the cells they interact with, forming handshakes that signal "this is self — do not attack."

Cancer learned to speak that language.

A tumour that upregulates PD-L1 on its surface can grab hold of PD-1 on approaching T cells and deliver the "stand down" signal. The T cell, receiving a message it was built to obey, disengages. It doesn't die; it exhausts, entering a state of immunological paralysis. The cancer grows undisturbed, wearing the body's own safety signals as camouflage.

Releasing the Brakes

The story of checkpoint inhibition begins with James Allison at the University of Texas, who spent years trying to convince a sceptical oncology community that blocking CTLA-4 — one of the first known immune brakes — could treat cancer. His drug, ipilimumab, was approved by the FDA in 2011 for metastatic melanoma. The field shifted.

What followed was a rapid expansion of targets. PD-1 and its ligand PD-L1 proved even more tractable. Nivolumab (anti-PD-1) and pembrolizumab (anti-PD-1) were approved in quick succession across an expanding list of cancers: melanoma, lung, kidney, bladder, head and neck, Hodgkin's lymphoma. Atezolizumab targeted PD-L1 directly, blocking the cancer-side of the handshake instead of the T cell side.

The results, in patients who respond, are unlike anything in oncology's history. Metastatic melanoma — which killed nearly everyone within a year in the 1970s — now has a substantial fraction of patients alive at five and ten years. Some appear cured, though oncologists use that word with characteristic caution. The response is not universal: roughly 20–40% of patients with many tumour types respond meaningfully to checkpoint blockade. The question of who responds, and why, is one of the most active areas of cancer research.

Mukherjee is careful to frame this uncertainty in human terms. He describes standing with medical students in the oncology ward, looking at a patient with metastatic melanoma, and explaining that she might be in the 15% who survive long-term — or she might not. There is, for now, no reliable way to know in advance. This is not a failure of the drugs; it is the frontier where the science currently lives.

The CAR-T Revolution

If checkpoint inhibitors remove the brakes, CAR-T therapy rebuilds the engine.

Chimeric antigen receptor T cells — CAR-T cells — are the patient's own immune cells, extracted, genetically reprogrammed in the laboratory to target a specific cancer antigen, expanded into billions, and reinfused. The work was pioneered by Carl June at the University of Pennsylvania and others who spent decades refining the viral delivery of the CAR gene, the expansion protocols, and the management of the extraordinary toxicities that can result when billions of armed T cells flood into a tumour-bearing body.

The clinical results in blood cancers — acute lymphoblastic leukaemia (ALL), diffuse large B-cell lymphoma, multiple myeloma — have been, in some patients, dramatic. Children with relapsed ALL, for whom every other option had failed, entered remission. Some stayed there. These are patients who, a generation ago, would have had no realistic prospect of survival.

Mukherjee is not a bystander to this work. He co-founded Immuneel Therapeutics in Bangalore in collaboration with Kiran Mazumdar-Shaw, with the explicit goal of conducting CAR-T clinical trials designed to reduce the cost of the therapy for patients in India and the developing world — where a treatment costing $400,000 per infusion is effectively nonexistent. His lab at Columbia is working on the next generation: solid tumour CAR-T, which remains far harder than blood cancer because solid tumours create hostile microenvironments that exhaust and exclude infiltrating T cells.

Cancer as a Successful Parasite

Mukherjee's framing of cancer immunotherapy carries his broader intellectual signature. He does not treat immunotherapy as a triumph to be celebrated cleanly. He treats it as a clarification — a new angle from which the fundamental problem of cancer becomes sharper and more demanding.

Cancer succeeds, he argues, because it is not alien. It is self. A cancer cell, at its origin, was a normal cell that accumulated the right (or wrong) combination of mutations to break free of the regulatory circuits governing growth, death, and identity. What it uses to survive — growth signals, evasion mechanisms, the ability to recruit blood supply and suppress immune recognition — is the same toolkit the body uses for legitimate purposes. The immune system, having evolved to distinguish self from non-self, is badly equipped to deal with an enemy that is genuinely both.

Immunotherapy works, when it works, by tilting that balance. Tumours accumulate mutations, and some of those mutations produce neoantigens — protein fragments that are sufficiently different from normal self that, if the immune system were paying full attention, it would recognise them. The checkpoint inhibitors restore that attention. They do not create a new immune response; they remove the suppression of one that was already there, waiting.

What Comes Next

The open problems are the ones Mukherjee finds most interesting, and most honest to describe.

Solid tumour CAR-T remains the hard frontier. Blood cancers work because T cells can circulate and reach their targets. Solid tumours — pancreatic cancer, glioblastoma, ovarian cancer — build barricades: a suppressive microenvironment that excludes and exhausts T cells before they reach the tumour core. Solving this requires not just better receptors but better understanding of how tumours co-opt the surrounding stroma, blood vessels, and immune architecture to protect themselves.

Combination therapies — checkpoint inhibitors plus chemotherapy, plus targeted therapy, plus CAR-T — are multiplying faster than our ability to understand their interactions. The signal in any individual combination trial is often confounded by the complexity of what the drugs are doing simultaneously.

Tumour heterogeneity compounds everything. A single tumour may contain many distinct cell populations, each with different mutation profiles, different antigen expression, different susceptibility to immune attack. A therapy that clears 99% of cells but leaves one resistant subclone will fail. This is not a theoretical concern; it is the pattern of relapse that oncologists observe constantly.

And cost. Mukherjee returns to this with the insistence of someone who has thought hard about who gets to benefit from a revolution. CAR-T therapy, at current prices, is available to a vanishingly small fraction of the world's cancer patients. Checkpoint inhibitors are cheaper but still out of reach for most. The science is advancing faster than the delivery mechanisms. Immuneel, his Indian venture, is a bet that manufacturing and trial costs can be reduced enough to change that equation — but it remains a bet.

The Long Arc

Mukherjee keeps a long lens on cancer history. He begins The Emperor of All Maladies with the Edwin Smith Papyrus, an Egyptian medical document from roughly 1600 BCE, which describes breast cancer and notes of it, plainly, that there is no treatment. He ends the book in a different register — not triumphalism, but the recognition that the arc of cancer medicine, measured across centuries, has been one of slow, hard-won, sometimes reversible progress.

Acute lymphoblastic leukaemia in children went from 100% mortality in 1947 to 90% survival today. Melanoma went from a near-universal death sentence to a disease with meaningful long-term survivors. The progress is real. The uncertainty about any individual patient remains profound.

What immunotherapy adds to that arc is a new and finally coherent theory: that cancer is not, at bottom, a problem of chemistry or radiation tolerance, but of recognition. The immune system is the body's most sophisticated surveillance apparatus. Cancer's greatest trick has been to become invisible to it. The task of the next generation of cancer medicine — checkpoint inhibitors, CAR-T, cancer vaccines, bispecific antibodies, tumour microenvironment engineering — is to make it visible again.

Synthesized from Siddhartha Mukherjee's lectures, interviews, and research — including NHGRI oral history, Columbia Grand Rounds, the Weizmann Global Gathering, and The Emperor of All Maladies, The Gene, and The Song of the Cell.