Mister LibraryThe Librarian's Regimen
Español

The First Blood Test for Fifty Cancers Reached the FDA Today. It Still Misses Most Early Ones

SEPTEMBER 23, 2026

A seven-panel 1882 scientific plate by Walther Flemming showing successive stages of a single cell dividing in two, with dark thread-like chromosomes pulling apart between star-shaped spindles
Walther Flemming's original 1882 drawings of mitosis — one cell, caught in the act of becoming two — from Zellsubstanz, Kern und Zelltheilung, the book that first described the process. Public domain, via Wikimedia Commons.

Can a blood test catch cancer before it's really cancer — at the first cell, rather than the billionth? That is the question behind a tweet that has been going around this week, promoting an episode of the podcast No Time But Now in which the oncologist Dr. Azra Raza tells William Shatner and Melanie Shatner Gretsch that medicine has spent decades chasing the last cancer cell with chemotherapy, radiation and surgery, when it should be hunting for the first one. It is a real question with a forty-year research career behind it. And, by the same week's coincidence, the technology that comes closest to answering it in the real world — Galleri, a blood test from the company GRAIL that screens for a signal shared by more than fifty cancer types from one blood draw — had its first-ever hearing in front of an FDA advisory panel today, September 23, 2026 [1] [2]. Here is what that test actually finds, what a 142,942-person randomised trial found it missing, and how much further Dr. Raza's own, considerably more radical idea still has to go.

The tweet in question:

Who Dr. Azra Raza is, and what "the first cell" means

Dr. Raza is the Chan Soon-Shiong Professor of Medicine and director of the Myelodysplastic Syndrome (MDS) Center at Columbia University, and an international authority on MDS — the group of pre-leukemic bone marrow disorders in which roughly a third of patients eventually progress to acute myeloid leukemia [3] [4]. Since 1984 she has banked blood and bone marrow from her own MDS and leukemia patients, building a tissue repository that now holds more than 60,000 samples across 35-plus years — reportedly the oldest of its kind assembled by a single physician — and her lab's work on the biology of MDS contributed to the approval of lenalidomide, a drug still used to treat it [3] [4]. Her 2019 book, The First Cell: And the Human Costs of Pursuing Cancer to the Last, makes the argument in the tweet at length: that oncology has become extraordinarily good at fighting cancer once it is advanced and genetically chaotic, and comparatively uninterested in the much harder, much less funded problem of finding the first abnormal cell while it is still one cell, or a few, rather than a tumor [5]. It is a credentialed argument, not a fringe one — which is exactly why it is worth checking what actually answers it today, separate from what a podcast tweet implies is imminent.

The technology closest to "the first cell" is already for sale

The general category is called multi-cancer early detection (MCED), and Galleri is the furthest along. The idea: a tumor sheds fragments of its own DNA into the bloodstream (cell-free DNA), and that DNA carries chemical tags called methylation patterns that differ, in a fairly consistent way, between healthy tissue and cancerous tissue — and between different types of cancer. Galleri sequences those patterns across the genome, uses a machine-learning classifier to flag a likely cancer signal, and, when it finds one, predicts which organ it came from (a "Cancer Signal Origin" call, right about 93.4% of the time it makes one) [6]. None of this is hypothetical or pending: GRAIL has sold Galleri since September 2021, as a physician-ordered laboratory test regulated under CLIA rather than as an FDA-approved product, starting in New York and expanding from there [7]. It is available right now, this afternoon, to anyone whose doctor will order it, for a list price of $949 (as low as $799 through some providers); insurance coverage is still limited, though TRICARE now covers it with prior authorization [6]. What changed this year is that GRAIL filed for full FDA Premarket Approval on January 29, 2026 [1], and today's session — the Molecular and Clinical Genetics Panel of the FDA's Medical Devices Advisory Committee — is the first time any FDA advisory panel has ever convened to review an MCED test [2]. So "when will this be widely available" has an odd answer: it already is, at a price most people pay out of pocket; what today's hearing is actually about is whether it earns broader insurance coverage and an FDA-sanctioned claim.

What the trials actually found

Two studies carry the weight of the application. PATHFINDER 2 followed 25,490 people for one year [1]. The much larger NHS-Galleri trial randomised 142,942 people aged 50 to 77 in England with no cancer symptoms into a screened group and a usual-care control group [9]. The screened group had four times as many cancers found through the trial itself as the control group (1,173 versus 290), 14% fewer diagnoses at stage IV, and 20% fewer cancers found the hard way, through an emergency-room presentation [9]. Those are real, meaningful numbers. But the trial's own prespecified primary endpoint — a statistically significant reduction in the proportion of cancers diagnosed at stage III or IV, across a defined list of 12 cancer types — was not met [9]. And the test's overall accuracy, by its own trial data, is modest: it correctly flagged 30.7% of the cancers actually present in screened participants, while giving a false alarm only 0.45% of the time (specificity around 99.6%) — meaning a positive result was confirmed as cancer just over half the time (52.0% positive predictive value) [8]. Broken out by stage, the pattern is stark: the FDA's own staff briefing, released ahead of today's panel, puts stage I detection at just 21.4% in PATHFINDER 2 and 13.6% in NHS-Galleri, against roughly 60% for stage IV disease [10] — weakest, in other words, at exactly the stage where catching cancer early matters most. FDA staff flagged no major safety issues and no adverse events attributed to the test itself, but they did raise a sharper concern: whether a test that misses roughly four in five stage I cancers should be marketed with the word "early" at all, and whether a negative result might give people false reassurance and lead them to skip mammograms, colonoscopies and the screening that already exists [10]. No trial has yet shown that using Galleri actually reduces cancer deaths — that mortality read-out from the UK trial is still years away [9].

What today's vote does and doesn't decide

The panel's recommendation is advisory, not binding — the FDA makes the final call, though it usually follows its panels' lead [9]. A favorable vote today would not, by itself, put Galleri in more doctors' hands tomorrow: full approval would follow separately, could carry a narrower label than GRAIL requested (describing "cancer detection" rather than "early detection," given the stage I numbers above), and broad Medicare reimbursement — the thing that would actually make this a mass-screening tool rather than an affluent early-adopter's $949 purchase — is realistically not expected before 2028 [9].

Dr. Raza's own, much longer shot

Galleri is not, as far as the public record shows, Dr. Raza's own project. Her own venture is considerably more ambitious and considerably earlier-stage. In a March 2025 interview with the biologist Michael Levin, she described a company, PredictRx, and a parallel effort called The Repository — a consortium including Columbia, Harvard and Johns Hopkins collecting blood, saliva, stool, urine, fingernail and hair samples from cancer patients in remission, aimed at what she calls "Stage Negative One": detection before a tumor exists at all [11]. The device she describes, called "Stentinel," is a stent implanted in the arm carrying electronic sensors, designed to continuously monitor the bloodstream for polyploid giant cancer cells (PGCCs) — unusually large, multi-nucleated cells some researchers believe can seed a cancer's return — and to flag one within 18 days of its appearing [11]. PGCCs themselves are a genuine, published area of cancer biology; a 2024 paper, for instance, describes circulating PGCCs as a potential prognostic marker in carcinoma patients [12]. But that is early mechanistic research, not proof that an implanted sensor can catch one the moment it appears. I could not find a company website, a peer-reviewed clinical study, or any FDA filing for Stentinel or PredictRx — the entire public record, at the time of writing, is a small number of interviews and Dr. Raza's own social media posts. That does not make the underlying biology wrong. It does mean this idea sits years, at an absolute minimum, behind Galleri, which already has two completed studies of tens of thousands of people and an actual FDA docket.

The sponsor, for the record

The podcast episode itself is "Presented by LifeWave" [15] — a company that sells adhesive skin patches marketed as a patented form of "phototherapy" claimed to boost stem cell activity, energy and sleep, sold through a multi-level-marketing distributor network [13]. No large independent trial supports those specific product claims, and in April 2026 the FTC settled charges against two of the company's senior distributors for deceptive income claims; the company's own disclosure, cited in that case, showed 79% of its active participants earned nothing in commissions in 2024 [14]. None of that reflects on Dr. Raza's own credentials, which stand independently on a public research and clinical record — but a reader clicking through from a sponsored podcast is entitled to know who paid for the microphone.

What the evidence says. The idea in the tweet — find cancer's first cell instead of chasing its last — already has a real, partial answer on the market: Galleri, a $949 blood test sold since 2021, which had its first FDA advisory hearing today. In its own 142,942-person randomised trial it found four times more cancers than usual care and cut stage IV diagnoses by 14%, but it missed its own prespecified primary endpoint, catches only 13.6–21.4% of stage I cancers against roughly 60% of stage IV, and has no mortality data yet. Dr. Raza's own, more radical version of the same idea — a sensor implanted in the arm, built to catch a single rogue cell within 18 days — exists today as a small venture and a handful of interviews, with no published trial anyone can check. Both are serious attempts at a forty-year-old problem. Only one of them is something a doctor can order this afternoon.

Sources

  1. GRAIL. GRAIL Submits FDA Premarket Approval Application for the Galleri Multi-Cancer Early Detection Test. Press release, 29 January 2026. grail.com
  2. U.S. Food and Drug Administration. September 23, 2026: Molecular and Clinical Genetics Panel of the Medical Devices Advisory Committee Meeting. fda.gov
  3. Herbert Irving Comprehensive Cancer Center, Columbia University. Azra Raza, MD — faculty profile. cancer.columbia.edu
  4. NewYork-Presbyterian. Taking Aim at Pre-Leukemia Disorders: NewYork-Presbyterian/Columbia Establishes Myelodysplastic Syndromes Center. nyp.org
  5. Raza, Azra. The First Cell: And the Human Costs of Pursuing Cancer to the Last. Basic Books, 2019.
  6. GRAIL. Frequently Asked Questions (FAQs) for Patients | Galleri. galleri.com
  7. GRAIL. GRAIL Receives New York State Approval for Galleri Multi-Cancer Early Detection Blood Test. Press release, 20 September 2021. grail.com
  8. NHS-Galleri Trial. How accurate was the test? nhs-galleri.org
  9. Medical Daily. One Blood Test, Two Verdicts: FDA Advisers Weigh Galleri Wednesday After Randomized Trial Missed Its Main Goal. September 2026. medicaldaily.com
  10. Medical Daily. Galleri Blood Test Draws No Major Safety Questions From FDA Staff, but It Missed Most Stage I Cancers. September 2026. medicaldaily.com
  11. Evolution 2.0. New Theory & Practice for Cancer — Azra Raza and Michael Levin. March 2025. evo2.org
  12. Circulating Polyploid Giant Cancer Cells, a Potential Prognostic Marker in Patients with Carcinoma. PMC11432346
  13. Office for Science and Society, McGill University. Patchy Science on LifeWave's Mysterious Patches. mcgill.ca
  14. Federal Trade Commission. FTC Takes Action Against High-Level MLM Participants Who Deceived Workers About the Amount of Money They Can Earn. Press release, 27 April 2026. ftc.gov
  15. No Time But Now (@notimebutnowpod). Tweet, 22 September 2026. x.com
cancercancer screeningAzra Razamulti-cancer early detectionGalleriGRAILliquid biopsyFDAmethylationmyelodysplastic syndromeoncology

This is one reader's reading of the research, not medical advice. If something here touches on your own health, take it to a clinician who knows you — and read how these entries are put together.

💬 Comment on X🔍 See what others said

Keep reading

SEPTEMBER 17, 2026 Pancreatic Cancer Has a Drug That Works. Now the FDA Wants It Tried First Daraxonrasib doubled survival in treated pancreatic cancer; now it has FDA Breakthrough status for first-line use with chemo. The trials and the company. SEPTEMBER 17, 2026 What Radioligand Therapy Is, and Why a Second Lutathera Matters What radioligand therapy is, what Lutathera’s trials showed, and why Curium’s Bexlutry — the first radioligand equivalent — matters, read from the studies. SEPTEMBER 23, 2026 Do Calcium and Vitamin D Supplements Actually Prevent Fractures and Falls? A BMJ review of 69 trials, 153,902 adults, found calcium and vitamin D don't meaningfully cut fracture or fall risk. SEPTEMBER 23, 2026 Folate vs. Folic Acid: What's the Difference, and What Does the Evidence Say About Autism? Folate and folic acid aren't identical. Here's the real difference, why it's fortified into bread, and what the autism research actually shows.

← Back to the Regimen