The First Cell: early detection and what screening can do

Azra Raza is a professor of medicine at Columbia. She directs its MDS Center. MDS stands for myelodysplastic syndromes, a group of bone marrow disorders that can turn into acute myeloid leukemia (AML).

The First Cell: And the Human Costs of Pursuing Cancer to the Last is not a history and not a memoir, though it contains both. It is a sustained argument, made through the deaths of specific people she treated.

A death stated plainly. Raza's husband died of leukemia. She writes about caring for him as both wife and doctor, and she does not hold it back as a reveal.

Her thesis is straightforward. Huge effort goes into treating cancer once it is advanced. For many cancers, the return on that effort has been small. Meanwhile almost nothing goes into finding the first cancer cell. That is the point at which the disease might be stopped. She wants the money, the talent and the attention moved.

The book is angry. The anger is aimed at a research system, not at individual doctors. She counts herself among the accused.

The book runs to 368 pages in the Basic Books edition, published 15 October 2019, with notes and references.

Its structure is unusual. Most chapters are built around a single patient, named or pseudonymous, followed from diagnosis to death. Raza gives the clinical course and the treatments tried. She gives what each cost the person in suffering, and what it bought in time. The build-up is deliberate. By the fifth or sixth chapter, the reader has stopped expecting a good outcome.

Between those chapters she makes the argument. She writes about the biology of MDS and how it turns into AML. She writes about why animal models have carried over so poorly into human benefit. She writes about trials that produce statistically significant results with little meaning for patients. And she writes about the money in drug development.

There is also a chapter on her husband's illness and death. It is the emotional centre of the book, and the hardest section to read.

The final part is constructive. She sketches what a research programme aimed at the first cell would look like. Tissue repositories. Sampling people at risk over time. And a shift in what counts as a question worth funding. She acknowledges the tools for it largely do not exist yet.

The case histories are extraordinary. Raza refuses the usual grammar of cancer writing. In that grammar, suffering is squeezed small and outcomes are described in the passive voice. She writes what mucositis looks like. What a failed transplant does to a family. What a fourth-line treatment costs a person in their last months.

She is also credible on her own specialty. MDS and AML are where her claim is strongest, and NCI's own summary supports part of the picture. MDS is reclassified as AML when blasts reach 20%. The acute leukemic phase is less responsive to chemotherapy than de novo AML. The only potentially curative option is allogeneic stem cell transplant. And in one trial in higher-risk disease, azacitidine gave a median overall survival of 24 months against 16 months for other regimens. Those are real numbers and they are small.

Her handling of animal models is another strength. She explains why a mouse cured of an implanted tumor tells you less than it seems to, and why the pipeline keeps producing drugs that work in that setting and not in people. It is a technical point made accessible without being softened.

The argument about incentives is worth taking seriously too. A research system that rewards small gains in advanced disease will produce small gains in advanced disease. Nobody planned it that way.

The generalisation from MDS and AML to cancer as a whole is the weak joint. NCI says Hodgkin lymphoma can usually be cured if found and treated early. It also says most testicular cancers can be cured, even if diagnosed at an advanced stage. Carrying the same gloom over to them would be wrong. The book largely leaves those cases out.

The early detection argument is also treated as more straightforward than it is. Finding cancer earlier is not automatically better. The book does not really take on lead-time bias, length bias or overdiagnosis. Those are the very reasons early detection sometimes fails to save lives.

Selection matters as well. Chapters built around patients who died produce a book in which patients die. That is an honest reflection of her clinic and a partial reflection of oncology.

Her proposals are also thinner than her diagnosis. Sampling healthy people over years, at scale, is costly and slow, and it raises hard ethical questions. The book waves at those problems rather than working through them. That is a fair limit for a book of this kind, but it means the constructive chapters carry less weight than the critical ones.

Finally, a reader in treatment should know what they are picking up. Nothing here is meant to reassure, and the descriptions of end-stage disease are graphic.

This is where the federal material adds what the book leaves out.

NCI's screening overview states the purpose plainly. Screening looks for cancer before a person has symptoms. Early detection may make a cancer easier to treat or cure. It then states the limit just as plainly. For some cancers, finding and treating the cancer early does not improve the chance of a cure. Nor does it help the person live longer.

NCI also says all tests carry possible risks. Problems from the procedure itself. False-positive results, which cause worry and more testing. And false-negative results, which can delay diagnosis if symptoms appear later.

NCI names some tests as proven. Among those shown to decrease the chance of dying from cancer are mammograms for breast cancer, and sigmoidoscopy and fecal occult blood testing for colorectal cancer. That list is short by design, because the standard is high.

NCI's discussion of screening statistics explains why. Lead-time bias makes survival look longer when a diagnosis is simply made earlier. Take a man diagnosed at 60 rather than 67, who dies at 70 either way. He has better five-year survival and not one extra second of life. Length bias means screening preferentially catches slow-growing cancers. Overdiagnosis is the extreme of that, and NCI cites estimates of 19% of screen-detected breast cancers and 20% to 50% of prostate cancers. NCI's conclusion is blunt. The only reliable way to know whether a screening test reduces deaths is a randomized trial showing a drop in cancer deaths.

That is not a refutation of Raza's argument. It is the specification her argument has to meet. For the details, see cancer screening overview and overdiagnosis.

The book was published just as multi-cancer early detection tests began to be sold, and readers often connect the two.

NCI's position, in a 2022 piece, is specific. FDA-cleared liquid biopsies exist for people who already have a cancer diagnosis. MCED tests meant to screen people without symptoms were still under study. NCI said randomized trials were needed.

Things have moved since then. The FDA briefing for a September 23, 2026 advisory panel meeting shows the agency was reviewing one MCED test, Galleri, for screening. A panel meeting is not an approval. Check the FDA for any final decision.

NCI lists the open questions. Who should be tested, and with which test. How the tests perform across different racial, ethnic and socioeconomic groups. Whether take-up would pull people away from screening that is already proven. And, most of all, whether these tests reduce cancer deaths at all. It names two specific harms. False positives lead to invasive and costly procedures that find nothing. And overdiagnosis picks up slow-growing cancers that would never have caused symptoms.

NCI said it was using expert feedback to guide the design of randomized trials to answer the death-rate question. It did not treat the tests as settled. Until a trial shows a drop in cancer deaths, a marketed MCED test is a research question being sold as a service.

So the honest version of Raza's argument is this. Finding cancer at the first cell would change everything, if the finding changed what happened next. No trial has yet shown, in NCI's words, that these tests reduce cancer deaths. That is a reason to fund the work, not a reason to buy the test. Raza's argument is that treating late disease absorbs the money early detection would need.

For where these tests stand in detail, see multi-cancer early detection blood tests.

It suits readers who want to know what advanced blood cancer actually involves, told plainly. It suits people interested in how research priorities get set. It suits clinicians and students, and it has been widely assigned for that reason.

It suits readers of Mukherjee and Kalanithi who want the other side of the case. It works against any book that treats cancer progress as a steady upward line.

It suits a reader who is already angry about how cancer is discussed and wants that anger given shape by someone qualified to hold it.

It does not suit someone in active treatment for a blood cancer, or a caregiver in the middle of one. The case histories are not written to be survivable reading in that position. It also does not suit a reader who wants a balanced account of cancer research. The book is an argument, and says so.

• Basic Books / Hachette — The First Cell
• Open Library — The First Cell (ISBN 9781541699519)
• NCI — Cancer Screening Overview (PDQ), Patient Version
• NCI — What Screening Statistics Mean
• NCI — Can Blood Tests Detect Cancer Early? What to Know About MCED Tests
• FDA — Executive Summary for the September 23, 2026 meeting of the Molecular and Clinical Genetics Panel (Galleri)
• NCI — Hodgkin Lymphoma Treatment (PDQ), Patient Version
• NCI — Testicular Cancer
• NCI — Myelodysplastic Syndromes Treatment (PDQ), Health Professional Version

This page discusses a book for education. It is not medical advice, and nothing here is a judgement of anyone's real medical care.