Daylila

Biotech & Longevity · Sunday, 6 September 2026

01 Briefing what happened

A mushroom compound given before chemotherapy prevented its worst lasting side effect. In mice.

Biotech & Longevity 50 sources

Most people on platinum chemotherapy end up with burning, numb hands and feet, and nothing treats it once it arrives. Two doses of psilocybin beforehand stopped it completely in mice, without weakening the chemotherapy, and human trials start this month.

up to 60%

of people given platinum chemotherapy who get lasting nerve damage

those drugs are standard treatment for ovarian and lung tumours [1]

2 doses

of psilocybin, given before each round of chemotherapy

the mice went through six rounds over eight months and kept normal paw sensation [1]

8 months

the entire length of the mouse study

the protection held for all of it, and nobody knows what happens after that [1][2]

The lead story — what happened

  • Researchers at the MD Anderson Cancer Center in Texas gave mice two doses of psilocybin before each of six rounds of chemotherapy. [1][2]
  • The treated mice kept normal feeling in their paws and kept the nerve endings in their skin, right through the eight months the study ran. [1]
  • The untreated mice became hypersensitive to pain, lost feeling in their paws, and their skin nerve fibres withered. [1]
  • The psilocybin did not blunt the chemotherapy. Tumours shrank as much as before and the immune response was unchanged. [1][2]
  • The condition is called chemotherapy-induced peripheral neuropathy: burning, tingling or numb hands and feet, sometimes for years after treatment ends. Nothing reliably prevents or undoes it. [2]
  • Chemotherapy causes it by stopping mitochondria, the power packs inside a cell, from travelling out to the far tips of sensory nerves. Starved of energy, the tips die back. [1]
  • Psilocybin switches on particular receptors on those nerve fibres. That sets off a chain of reactions ending in motor proteins that keep the mitochondria moving. [1]
  • The trip is not the active ingredient. A compound called tabernanthalog, which hits the same receptors without causing hallucinations, protected the mice about as well. [1]
  • This is a mouse study. Human trials begin this month, with the drug given as a pill. [2]
  • The paper was published in the journal Science on 3 September. [3]

Who is involved

  • Moran Amit

    a surgical cancer doctor at the MD Anderson Cancer Center in Texas and a co-author; he says the striking part is preventing the pain rather than treating it [1]

  • MD Anderson Cancer Center

    the Texas hospital and research centre whose team ran the mouse work [2]

  • Joe Cichon

    an anaesthesia researcher at the University of Pennsylvania who was not part of the study; he calls giving the drug first a clever approach, and says the mitochondria finding surprised him [1]

  • Science

    the journal that published the paper on 3 September [3]

What is pushing on this

Pressure to prevent rather than repair Building

there is no effective treatment once the nerve damage has set in [2]

Distance from mice to people High

every result here is in animals; the first human trial starts this month [2]

Caution about psychedelic drugs Easing

New Zealand approved MDMA for severe post-traumatic stress disorder the next day [9]

How it unfolded

  1. Before each round the mice get two doses of psilocybin [1]
  2. Six rounds, eight months treated mice keep paw sensation and nerve endings; untreated mice lose both [1]
  3. 3 September the paper is published in Science [3]
  4. This month the first human trials begin [2]

Where this points

Watch whether the human trial can be run the same way: it means giving a psychedelic to cancer patients before chemotherapy, then waiting to count nerve damage that never appears. [2]

The rest of the day

45 more stories on this beat.

Each with its own sources. None of these is a link to the story above.

  1. 02

    Breast cancer pill cleared for an earlier moment

    The US drug regulator gave accelerated approval on 4 September to Etcamah, also called camizestrant, for advanced breast cancer that is hormone-receptor positive and HER2 negative. [4] It is taken alongside one of three existing drugs. What is new is the timing: it starts the moment a blood test finds an ESR1 mutation, which tumours pick up to escape the standard first treatment. Fewer than 5% of patients have that mutation at diagnosis. Nearly 40% have it once the first treatment stops working. [4]

    Why it matters — The regulator said plainly that nobody has yet confirmed whether starting at the blood test, rather than when the cancer is visibly growing, does patients any good. It has ordered further studies to find out. [4]

  2. 03

    Two gentler drugs beat full chemotherapy in leukaemia

    Adults newly diagnosed with acute myeloid leukaemia, a fast-moving blood cancer, are normally given intensive chemotherapy if they are fit enough for it. In a trial of 172 such adults, half got that and half got azacitidine plus venetoclax, two milder drugs previously kept for people too frail for chemotherapy. [5] The median time before the disease came back or the patient died was 14.5 months on the two drugs, against 6.2 months on chemotherapy. Severe infections hit 28% against 41%. [5]

    Why it matters — It was a phase 2 trial of 172 people, funded by AbbVie, which sells venetoclax. That is a real result and one paid for by a company with a stake in it. [5]

  3. 04

    AbbVie reports myeloma results

    AbbVie reported results from a late-stage trial called Cervino of etentamig. The drug grips a cancer cell with one arm and one of the patient's own immune cells with the other, holding the immune cell against the target. [6] The 393 patients had multiple myeloma, a cancer of the bone marrow, and had already been through at least two rounds of treatment covering all three standard drug types. More than 130,000 people worldwide are diagnosed with it each year. [6]

    Why it matters — Myeloma has shifted from a quick death to a long illness, but it remains almost always incurable and nearly everyone relapses. This is a drug for the point where the standard three types have all stopped working. [6]

  4. 05

    GSK pays for a Chinese cancer drug

    GSK, a British drugmaker, agreed on 3 September to pay Hutchmed of China $110 million up front, and up to $1.19 billion more later, for most of the rights to a cancer drug called HMPL-A830. [7] Drugs of this family normally tie a homing antibody to a poison. This one ties the antibody to a small molecule that blocks KRAS, one of the commonest faults in human tumours. Hutchmed says laboratory tests showed the two halves working better together than either alone. [7]

    Why it matters — Almost all the money is conditional. GSK is paying $110 million for a drug that has never been given to a person, and the rest only if it survives the trials. [7]

  5. 06

    Immune drug added to breast cancer chemotherapy misses

    Doctors gave 773 women with triple-negative breast cancer, an aggressive form with no hormone target to aim at, an immune drug called atezolizumab on top of standard chemotherapy before surgery. Another 777 got a dummy. [8] After four years, 3.3 percentage points more of the atezolizumab group were alive with no return of the cancer. The trial's own statistics say that gap could be chance. Survival differed by 0.7 points, and serious side effects were 75.3% against 73.4%. [8]

    Why it matters — A planned look at subgroups suggested the drug did help women whose cancer had reached the lymph nodes. A subgroup finding inside a trial that missed its main goal is a lead for the next trial, not a result. [8]

  6. 07

    New Zealand approves MDMA for severe PTSD

    New Zealand's medicines regulator, Medsafe, approved pharmaceutical-grade MDMA on 4 September for people with severe post-traumatic stress disorder, a lasting condition that can follow a terrifying event and that resists the usual treatments. [9] Two psychiatrists won the right to prescribe it alongside talking therapy. MDMA is best known as a party drug. New Zealand is only the second country to allow this, after Australia in 2023, and the same government backed magic mushrooms for severe depression last year. [9]

    Why it matters — The approval went to named prescribers rather than onto the general list, so the number of people who can actually be given it stays very small. [9]

  7. 08

    US regulator tells overseas trial sites the rules still apply

    Four directors at the US Food and Drug Administration published a joint statement on 3 September restating that Good Clinical Practice applies wherever a drug trial is run. [10] That is the rulebook covering consent, record-keeping and the safety of the people in a trial. Drug companies increasingly run trials in countries where patients are easier to recruit and costs are lower, and the agency inspects those sites. The heads of its drug, biologics, device and cancer centres all signed it. [10]

    Why it matters — A regulator does not usually announce that its existing rules still apply unless its inspectors have been finding places where they were not followed. [10]

  8. 09

    AI medical devices reach patients before authorisation

    The US drug regulator has accepted four digital health products into a pilot called TEMPO that lets them be sold before they have marketing authorisation. Products from two companies, Cadence and Limbic, are among them. [11] These are devices that use generative artificial intelligence to make decisions about a patient's care, and the agency has not settled how to judge them. The pilot feeds a Medicare experiment that pays for technology helping older Americans manage long-term conditions. [11]

    Why it matters — The agency is using the real world as the trial. The products reach patients first, and the evidence about whether they help is gathered afterwards. [11]

  9. 10

    Action against unproven stem cell clinics continues

    Paul Knoepfler, a stem cell scientist who tracks the industry, wrote on 3 September that the US drug regulator has kept sending warning letters to clinics selling unproven stem cell treatments, which he had expected to stop. [12] His worry was that the US health secretary, Robert F. Kennedy Jr., said on a podcast in 2025 that he had taken such treatments himself in the Caribbean. The best available count puts more than 2,000 such firms in the United States. [12]

    Why it matters — It is a rare case of an agency continuing to act against a business its own political leadership has personally used and praised. [12]

  10. 11

    Pentagon and NIH sign a biodefence deal

    The US Department of Defense and the National Institutes of Health, the country's main funder of medical research, have signed an agreement to work together on biodefence and preparing for pandemics. [13] The New York Times obtained a copy; it covers chemical and biological defence research. Democratic members of Congress objected, warning that the arrangement could let the military reach billions of dollars that Congress had set aside for infectious disease work. [13]

    Why it matters — Moving money from the purpose it was voted for to another one is normally a decision taken in public by a vote. This one was taken by two agencies signing a document. [13]

  11. 12

    A fight over what counts as a measles death

    Two former senior officials at the US Centers for Disease Control and Prevention wrote on 4 September about a dispute over the death count. Debra Houry and Demetre Daskalakis say the health secretary and the agency's current leadership are arguing with Pennsylvania about whether deaths there count as measles deaths. [14] States report notifiable diseases using case definitions agreed slowly by epidemiologists through a joint council with the agency. The authors say that process is now under threat. [14]

    Why it matters — If states and the national agency stop agreeing on what a case is, the national count stops counting anything, and each state's figure becomes its own. [14]

  12. 13

    A quarter of Birmingham five-year-olds miss MMR

    More than a quarter of five-year-olds starting school in Birmingham this autumn have not had both doses of the MMR vaccine, which protects against measles, mumps and rubella, according to figures from the UK Health Security Agency. [15] The city is bottom in the West Midlands at 74.1%, against 90.4% in Shropshire. Eight in every hundred five-year-olds there have had no vaccinations at all. A single MMR dose already gives up to 93% protection. [15]

    Why it matters — It is the lowest rate in the region, published in the same week that the United States is arguing over how to count the measles deaths it already has. [14][15]

  13. 14

    Ada Yonath has died at 87

    Ada Yonath, the Israeli structural biologist who first worked out how to map the ribosome, has died aged 87. [16][17] The ribosome is the machine inside every living cell that reads genetic instructions and builds proteins from them. Yonath found a way to grow crystals of it good enough for X-ray analysis, which most of her field had thought impossible. She shared the 2009 Nobel Prize in Chemistry for it, and was the fourth woman ever to win that prize. [16]

    Why it matters — Her later work showed how antibiotics kill bacteria by jamming their ribosomes, and that is the basis on which new antibiotics are now designed. [16]

  14. 15

    Mutating a whole genome shows what AI cannot predict

    Researchers led by Ben Lehner at the Wellcome Sanger Institute in England changed almost every single letter of the genome of phi X, a virus that infects bacteria, and recorded what each change did. [18] Phi X has 5,386 letters and 11 genes, and in the 1970s it was the first genome ever fully read. Most of the changes harmed the virus, but the team cannot explain why a quarter of the lethal ones kill it. [18]

    Why it matters — Artificial intelligence tools built to spot harmful mutations did badly here, on the most studied genome there is. The finding is that these tools need more real experiments, not more computing. [18]

  15. 16

    A measuring error inflated gene-silencing results

    Scientists using RNA-targeting CRISPR, a version of gene editing that cuts a cell's messenger copies instead of its DNA, normally check how much they silenced with a laboratory method called RT-qPCR. [19] A paper in Nature Biotechnology on 1 September found that the guide molecules steering the cut survive the extraction step and jam that measurement. The silencing then looks stronger than it was, across every CRISPR system the authors tested. [19]

    Why it matters — The authors recommend a different enzyme and a second independent method. That leaves figures measured the old way open to question, and there are a lot of them. [19]

  16. 17

    An eight-letter genetic alphabet is read

    All known life writes its genes with four letters. Researchers at the University of California San Diego showed that RNA polymerase, the enzyme that reads DNA and copies it into RNA, can accurately read an eight-letter alphabet containing four synthetic additions. [20] Detailed imaging showed the enzyme handling the artificial letters much as it handles natural ones. That suggests cells could run an expanded genetic system using machinery they already have. [20]

    Why it matters — The point of extra letters is to let engineered cells build things natural chemistry cannot spell. The barrier has always been whether the cell's own machinery would read them. [20]

  17. 18

    A ribosome that reads its own body

    A ribosome normally bumps into a messenger molecule at random and builds whatever protein that message describes. Researchers reported in Nature on 2 September that they had written a protein's instructions into the ribosome's own RNA, so the ribosome makes that one protein and nothing else. [21] They showed bacterial ribosomes doing it. Beyond making a single product at scale, it bears on how protein-making could have worked before cells were organised. [21]

    Why it matters — Biotechnology mostly makes proteins in vats of bacteria, and the yield depends on ribosomes finding the right message among all the others. This removes the search. [21]

  18. 19

    Closing in on what triggers long COVID

    Nature published a long piece on 2 September on how near researchers are to explaining why some infections leave people ill for years. A 2024 study put the number of people worldwide with long COVID at 400 million. [22] The article follows Amy Proal, who caught an infection in 2004, was told her blood work was fine, was sent to a psychiatrist, and was later diagnosed with ME/CFS, a condition of severe exhaustion after infection. She now runs a research foundation. [22]

    Why it matters — Illness after a virus was dismissed for decades precisely because the standard tests came back normal. The current search is for the trigger those tests miss. [22]

  19. 20

    Genes linked to the Big Five personality traits

    A very large study published in Nature on 2 September linked genetic differences to the Big Five personality traits, the standard psychological set that includes how outgoing and how conscientious a person is. [23] The links found are small. They were, though, less disturbed by family background than genetic links to education or income are. The team then connected the traits to taking up smoking, to COVID-19 infection risk, and to whether a person joins a research study at all. [23]

    Why it matters — That last one matters for every other genetic study. If agreeable and open people are likelier to volunteer, every volunteer database is tilted before anyone measures anything in it. [23]

  20. 21

    The paperwork went, the fear stayed

    In February 2025 the US drug regulator dropped the elaborate blood-monitoring programme that had governed clozapine for more than 30 years. Clozapine is the only approved drug for schizophrenia that has not responded to anything else, roughly one person in three with the illness, and the only one shown to cut suicide. [24] A child psychiatrist wrote on 4 September that removing the paperwork has not removed the caution that kept doctors from prescribing it. [24]

    Why it matters — The monitoring existed because clozapine can destroy a type of white blood cell. The writer's point is that the rule was blamed for a reluctance that has outlived it. [24]

  21. 22

    Nine cancer drugs launched above $400,000 a year

    Nine cancer drugs approved in 2024 came to market at more than $400,000 a year, according to a tally by the Institute for Clinical and Economic Review, a group that weighs what a medicine does against what it costs. [25] One CAR-T therapy now lists at $600,000 for a single infusion and a rival at $550,000. A daily pill for gut cancer is $520,000 a year. Two decades ago the prospect of a $100,000 cancer drug drew fierce criticism. [25]

    Why it matters — The bills land on Medicare, on employers, on patients and on everyone paying US taxes and insurance premiums, and nothing caps them beyond what a company thinks it can charge. [25]

  22. 23

    An argument that drug testing is the bottleneck

    An opinion piece in the New York Times on 4 September, by a biotechnology editor at the magazine Works in Progress, argued that the machinery for testing new drugs is now what holds cures back. [26] It opens with Annette Harlow, whose lung cancer treatment stopped working the day after Christmas in 2019 and who was told there was nothing left. A message the next day offered her a phase 1 trial in Houston of sotorasib, aimed at a gene fault called KRAS that scientists had long thought untouchable. She was among the first people in the world to take it, and was alive more than six years later. [26]

    Why it matters — The case rests on one woman who reached a trial through a message that happened to arrive. The argument is that this should be the system rather than the exception. [26]

  23. 24

    After the Angelman failure, researchers defend the approach

    Ultragenyx's drug for Angelman syndrome, a genetic condition causing severe developmental delay, failed its late-stage trial this week. [27] Researchers told STAT on 4 September that the miss does not condemn the whole idea. Mark Zylka, an Angelman researcher at the University of North Carolina, said it says nothing about the other trials still running. One failed trial, he added, does not mean 'this is a bad mechanism'. Other companies are still testing similar genetic medicines. [27]

    Why it matters — After a failure, whether the idea died along with the drug is a separate question, and it is usually left unasked. Here the researchers closest to it say it did not. [27]

  24. 25

    Nobody has bought Abivax

    Abivax, a French drugmaker, has strong late-stage results for obefazimod, a pill for ulcerative colitis, a disease that inflames and ulcerates the large intestine. Investors spent most of 2026 expecting somebody to buy the company for around $20 billion. [28] Nine months in, nobody has. Adam Feuerstein, STAT's biotech columnist, wrote on 3 September that takeovers across the industry are running unusually hot, which makes the absence of this one harder to explain. [28]

    Why it matters — The people who bought the shares expecting a sale are the ones carrying the wait, and the article says they are unhappy about it. [28]

  25. 26

    Roche bets on weight loss that keeps muscle

    Roche, a Swiss drugmaker, told Fierce Biotech on 3 September it is excited about a drug licensed from Hanmi of South Korea that acts on a receptor called CRF2, rather than copying the gut hormone that Ozempic-style drugs imitate. [29] Researchers have known for decades that switching that receptor on protects animals from wasting muscle. Roche's pitch is weight loss that keeps lean muscle. All the evidence so far is in animals, and Hanmi is running the first human trial. [29]

    Why it matters — Losing muscle along with the fat is the main complaint about the current weight-loss drugs, and every large drugmaker is now hunting a version that does not. [29]

  26. 27

    An AI-designed drug for rare obesity

    Superluminal Medicines raised money for drugs that switch on G-protein-coupled receptors, a family of proteins many existing medicines already target. [30] Its first candidate aims at MC4R, a receptor that sets appetite and how much energy the body burns, and should reach human testing by the end of the year. Rhythm Pharmaceuticals already sells an MC4R drug, Imcivree, for rare inherited forms of obesity. Superluminal is starting with Bardet-Biedl syndrome and obesity caused by damage to the hypothalamus. [30]

    Why it matters — It is aiming first at rare inherited obesity, where the fault is known and the patient group is small, before going anywhere near the general market. [30]

  27. 28

    An injectable scaffold for stroke damage

    Biomedical engineers at Duke University in the United States made a material that can be injected into the cavity left after an ischaemic stroke, where blood flow was cut off and tissue died. [31] In mice, the material drew in the animals' own immune cells, encouraged new blood vessels, supported nerve tissue and improved movement. It appears to work partly by turning neutrophils, immune cells usually associated with damage, into helpful ones. The work is published in Cell Biomaterials. [31]

    Why it matters — Stroke care is almost entirely about the first hours. This aims at the months afterwards, when the damage is settled and very little is offered. [31]

  28. 29

    A gel that keeps arthritis drugs in the joint

    Researchers at the University at Buffalo in the United States made a gel that is injected as a liquid into an arthritic joint and turns into a slippery semi-solid at body temperature. [32] It stays in place for several weeks, slowly releasing drug-carrying particles, and lubricates the joint while it is there. Osteoarthritis is the wearing away of the cartilage that cushions a joint. Injected drugs normally wash out fast, which is why they have to be repeated. [32]

    Why it matters — Keeping the drug inside the joint is also about keeping it out of the rest of the body, which is what limits how strong a dose can be given. [32]

  29. 30

    Omega-3 and aspirin match antibiotics on gum disease

    Scientists in Brazil ran a year-long trial in people with severe periodontitis, a gum infection that destroys the bone holding teeth in place. [33] One group got the standard antibiotics. The other got omega-3 with low-dose aspirin. About 58% of each group reached the treatment goal. The work came from teams at Albert Einstein Israelite Hospital, Guarulhos University, the University of Taubate and the Ribeirao Preto dental school at the University of Sao Paulo. [33]

    Why it matters — Resistance grows with every course of antibiotics prescribed anywhere, so a treatment that matches them without being one is worth more than the numbers alone suggest. [33]

  30. 31

    Sweetener effects showed up in later generations of mice

    Researchers fed mice sucralose and stevia, two zero-calorie sweeteners used in diet drinks. [34] Both changed the animals' gut bacteria, cut the useful compounds those bacteria produce, and altered the activity of genes involved in handling energy and inflammation. Some of those changes then appeared in later generations of mice that had never eaten either sweetener themselves. [34]

    Why it matters — It is a mouse study and the same question has not been put to people. What makes it unusual is the claim that a change carried into animals that never touched the stuff. [34]

  31. 32

    A berry compound cleared fat from muscle cells

    Scientists in Japan tested pterostilbene, a compound found in blueberries and grapes, on cultured mouse muscle cells. [35] It cut the abnormal fat that had built up inside those cells, both by speeding up fat breakdown and by stabilising a protein involved in handling fatty acids. Fat stored inside muscle cells, rather than under the skin, gets in the way of how muscle works and makes it harder for the body to use sugar properly. [35]

    Why it matters — This is cells in a dish, not an animal and not a person. Eating blueberries is not the same as dosing muscle cells with one of their compounds. [35]

  32. 33

    Dogs have a Lyme vaccine, people do not

    There is no licensed Lyme disease vaccine for people in the United States, although dogs have had one for years. [36] Lyme is a bacterial infection carried by ticks, and untreated it can leave people with fatigue, aches and nerve damage for years afterwards. The US disease agency estimates about 476,000 cases are diagnosed each year, and ticks are spreading into territory they did not previously occupy. Human vaccines are back in trials. [36]

    Why it matters — A vaccine has to be given to healthy people living where the ticks are, and then judged on infections that do not happen. That is why the human version has taken so much longer than the dog one. [36]

  33. 34

    Higher semaglutide doses tracked with fewer mental health events

    Researchers followed 63,215 patients who already had a neurological or psychiatric condition and then started a diabetes drug, tracking 24 different outcomes. [37] After matching patients on their other characteristics, those on semaglutide, the drug in Ozempic and Wegovy, had broadly fewer such events over two years than those on metformin or two other drug classes. Within the semaglutide group, a higher dose in the first two years went with fewer events in the next two. [37]

    Why it matters — Nobody was assigned a drug at random here, so the people who ended up on higher doses may differ from the rest in ways the matching did not catch. [37]

  34. 35

    Regulators jointly clear a substitute for crab blood

    Drug regulators from several countries, working through a joint body called ICMRA, completed a shared assessment on 31 August of a new way to test biological medicines for contamination. [38] The test in use relies on blood from horseshoe crabs, which are caught, bled and returned to the sea. The new approach uses a synthetic substitute. Having several regulators review it together means a manufacturer does not have to convince each one separately. [38]

    Why it matters — The joint review is the part that makes a switch possible. It has been the paperwork, not the chemistry, keeping the crab test in place. [38]

  35. 36

    Fauci gives his first long interview in a year

    Anthony Fauci ran the US National Institute of Allergy and Infectious Diseases for decades and became the face of the country's COVID-19 response. He gave a nearly two-hour public interview on 4 September at Georgetown University, where he has worked since 2023. [39] He had spent most of the past year out of view, and in July he repeatedly invoked his right not to answer at a Senate hearing. He defended his pandemic decisions, saying he is not sure much would change. [39]

    Why it matters — Whether the pandemic response was right is now being settled in hearings and interviews rather than in studies, and this was his first long answer in a year. [39]

  36. 37

    A catalogue of the microbes that live inside other things

    Most microbes cannot be grown in a laboratory, so what they do is largely inferred. Researchers used machine learning to sort more than a hundred thousand microbial genomes taken from environmental samples, predicting which of them live inside or on another organism. [40] Their estimate is that 15 to 23% of uncultivated microbes are symbionts, and that they turn up in half of all known bacterial and archaeal groups. The predictions are published as a catalogue called Symbiont Genomes. [40]

    Why it matters — These are predictions from a model rather than observations, so the catalogue's worth depends on somebody later confirming a sample of them in a laboratory. [40]

  37. 38

    Blood and spinal-fluid markers for Parkinson's stages

    Parkinson's disease damages the brain for years before tremor and stiffness appear, which makes it hard to diagnose early or to track. Researchers reanalysed protein and chemical measurements from the spinal fluid and blood of more than 1,100 people in a long-running Parkinson's study. [41] Machine learning picked out 21 candidate markers, and the best models correctly identified 83 to 86% of cases. Eight of the markers tracked how far the disease had progressed. [41]

    Why it matters — A test that works before symptoms start is only useful if there is something to do about it, and for Parkinson's there is not yet. [41]

  38. 39

    Three drugs older adults may be taking too readily

    The New York Times set out on 5 September a pattern in prescribing for older patients. Researchers find that a widely used drug is less effective or riskier in old age. More studies confirm it. Medical bodies eventually revise their guidance, and the drug may be added to the Beers Criteria, an influential American list of medicines that are often inappropriate for older adults. [42] Then the prescribing carries on much as before. [42]

    Why it matters — The evidence, the guideline and the prescription pad move at three different speeds. A patient gets whichever one their own doctor is working from. [42]

  39. 40

    What lockdowns did to the heaviest children

    The BBC reported on 3 September on the UK's Complications from Excess Weight clinics, a network for the most severely obese children, which has treated more than 6,000. [43] One of them, Ibbie from Doncaster, weighed nearly 22 stone, about 140kg, by the age of 12 and was told she might need a liver transplant if she did not lose weight. Her weight rose sharply during the COVID-19 lockdowns, when she stopped going to school entirely after being bullied. [43]

    Why it matters — The clinics exist for children whose illness has already arrived, and the story staff keep hearing starts with a routine that broke rather than with anything a family decided. [43]

  40. 41

    Two papers on what AI in drug hunting can do

    Two papers in the Nature journals on 1 September dealt with what artificial intelligence can and cannot do in drug research. One describes AdaptiveFlow, a free platform for searching enormous libraries of possible drug molecules; it comes with a ready-to-use version of a 69-billion-compound catalogue and cuts the computing bill by looking at the most promising chemical regions first. [44] The other argues the field still has no honest way to test whether large biology models work beyond the data they were trained on. [45]

    Why it matters — Searching more molecules and knowing whether the search means anything are separate problems, and only the first one has been solved. [44][45]

  41. 42

    A picture of how complex cells may have begun

    Researchers found a new microbe living inside stromatolites, the layered mats built by microbes that are among the oldest structures life has left behind. [46] They captured the first direct images of an Asgard archaeon physically joined to a bacterium, with the two apparently passing nutrients and other compounds between them. Partnerships like that, billions of years ago, are one leading explanation for how the first complex cells came about. [46]

    Why it matters — The idea that complex cells began as two simpler ones stuck together has been argued from genetics for decades. This is a photograph of two organisms doing it now. [46]

  42. 43

    Weekend eating times and heart disease

    Researchers used the French NutriNet-Sante study, which followed 104,806 adults from 2009 to 2023, to test whether eating at different times on work days and free days is linked to heart disease. [47] They measured the gap between weekday and weekend timing of the first and last calories of the day, taken from repeated 24-hour food diaries, and averaged 5.8 diaries per person. Heart disease was self-reported and then checked against medical records. [47]

    Why it matters — These are people who volunteered and wrote down their own meals, so the study can show a pattern and cannot show that shifting mealtimes caused anything. [47]

  43. 44

    Gene insertion made two to four times better

    Putting a whole new gene into a chosen spot in a genome is much harder than snipping one out. Researchers redesigned the landing sites that an enzyme called Bxb1 recognises and reported on 2 September that one engineered site reached 51.9% insertion efficiency in human cells, 1.7 times the natural version, and 35.6% in rice cells, 4.4 times. [48] With a matching change to the enzyme they inserted a CAR cancer-therapy cassette at 31%. [48]

    Why it matters — Cell therapies depend on getting a large piece of DNA into the right place, and this step decides how many cells in a batch actually end up carrying it. [48]

  44. 45

    Designed carriers built to do a virus's job

    Viruses spent billions of years getting good at one thing: pushing genetic material into cells. Researchers reported in Nature on 2 September that they had used protein-design software to build carriers from scratch that package and transfer RNA, instead of starting from a natural virus and stripping it down. [49] Most viruses have converged on similar shell sizes and shapes, and building from nothing lets researchers choose those properties rather than inherit them. [49]

    Why it matters — Every gene therapy has a delivery problem, and the usual answer is a modified virus that a patient's immune system may already recognise and attack. [49]

  45. 46

    Lab heart muscle grown with nerves attached

    Heart muscle grown in a laboratory usually has no nerves, which leaves out the part of the nervous system that drives the heart faster under stress. Researchers fused clusters of sympathetic neurons with engineered human heart muscle and showed the nerves forming working junctions with the muscle cells. [50] Using the model for long-QT syndrome type 2, an inherited rhythm disorder, they found the nerves themselves becoming overactive. [50]

    Why it matters — The fault was assumed to sit in the heart muscle's own ion channels. In this model the nerve is doing part of the damage, which points at a different target. [50]

02 Lesson why it matters

Why preventing a disease is harder to prove than curing one

A cure shows itself in a sick person getting better, but prevention can only be proved by counting an illness that never arrives.

The twist

A drug given to a healthy person has to clear a higher bar than one given to a dying person, because the healthy person had something to lose.

How it works

  1. A cure is given to people who are already ill
  2. You can watch each one get better, or not
  3. Prevention is given to people who are well
  4. Most of them were never going to get ill anyway
  5. So the difference only shows in large numbers over long periods
  6. And the drug must be safe enough for everyone who gains nothing from it

The same force, elsewhere today

Where this chain is also running, in today's other stories.

  • The breast cancer pill cleared for an earlier moment

    the same step, one stage back: the US regulator approved camizestrant for the moment a blood test spots a resistance mutation, before the cancer is seen to grow, and said nobody has yet proved that treating then helps

  • Dogs have a Lyme vaccine, people do not

    to prove the human version works, a trial has to give the shot to healthy people living where the ticks are and then count the infections that never arrive

  • A quarter of Birmingham five-year-olds miss MMR

    prevention that already worked is invisible, so the case for the missing quarter has to be made about a disease most of their parents have never seen

  • Blood and spinal-fluid markers for Parkinson's stages

    the same chain stops one step earlier: the markers can spot the disease before the symptoms, and there is nothing yet to give the person they find

Where you've seen this

Flood walls

you can count the towns that flooded, never the ones that did not

Car recalls

a fault fixed before anyone crashes leaves no crash to point at

Computer security

the break-in a patch prevented appears in no report anywhere

The catch

Waiting for proof is not free either. Chemotherapy nerve damage that has already set in has no treatment at all, so the people in this month's trial are the ones nobody can help later.

And the whole of it

Most of the medicine a person takes over a lifetime is preventive: a vaccine, a blood pressure pill, a statin, taken for years against an event that may never have been coming. Nobody swallowing one can tell whether it worked for them, and neither can the doctor who prescribed it.

03 Truth what's really going on

What is really going on

The most promising results on this beat today are all about acting before a person is ill. Psilocybin is given before chemotherapy [1], a breast cancer pill is started the moment a blood test finds a mutation [4], a Lyme vaccine is back in trials [36], and a quarter of Birmingham's five-year-olds have not had both MMR doses [15]. In every case the money and the risk are taken now, and the proof is years away.

Why it works on us — A person who was saved has a name and a face. The New York Times opinion piece has Annette Harlow, alive more than six years after a phase 1 trial she reached by chance [26]. Nobody can name the person who never got sick.

Who gains

  • AstraZeneca — Its breast cancer pill camizestrant is approved for the moment a blood test finds an ESR1 mutation. Fewer than 5% of patients have that mutation at diagnosis and nearly 40% have it after the first treatment fails, so the approval moves treatment earlier and lengthens it. [4]
  • AbbVie — It paid for the leukaemia trial that found azacitidine plus venetoclax beating intensive chemotherapy, and it sells venetoclax. [5]
  • Hutchmed — GSK is paying $110 million up front for a cancer drug that has never been given to a person, and carrying the cost of the trials that follow. [7]
  • Cadence and Limbic — Their generative-AI products can be sold to patients under the US regulator's TEMPO pilot without marketing authorisation, and feed a Medicare payment experiment. [11]
  • Makers of non-hallucinogenic psychedelic analogues — Tabernanthalog protected the mice about as well as psilocybin, which means the protective effect can be sold without the trip or the supervision a trip requires. [1]
  • The US Department of Defense — The agreement with the National Institutes of Health gives it a role in biodefence and pandemic work, which members of Congress say could reach money set aside for infectious disease. [13]

Who pays

  • People who already have chemotherapy nerve damage — The psilocybin worked only when given before the chemotherapy. Nothing in the paper repairs damage that has already happened, and nothing else does either. [1][2]
  • Children in Birmingham — More than a quarter of five-year-olds starting school there have not had both MMR doses, and eight in a hundred have had no vaccinations at all. [15]
  • Medicare, US employers and anyone paying US insurance premiums — Nine cancer drugs launched above $400,000 a year in 2024, and no rule caps the price beyond what a company thinks it can charge. [25]
  • Laboratories that measured gene silencing with RT-qPCR — The guide molecules jam the measurement, so their published silencing figures are overstated by an unknown amount and the work has to be redone with a different enzyme. [19]
  • Ultragenyx and families living with Angelman syndrome — The late-stage trial missed, and the researchers defending the approach are pointing at other companies' trials, not at a next step for this drug. [27]
  • States reporting disease numbers to the US disease agency — If the agreed case definitions stop holding, a state's own count becomes the only number it has, and its dispute with the national agency becomes public each time. [14]

What nobody knows yet

Open questions from across today’s stories — ours included.

  • 01

    How many people getting chemotherapy end up with lasting nerve damage.

    Nature puts it at up to 60% of those on platinum-based drugs; the New York Times says about two thirds of everyone who receives chemotherapy. Neither figure can be checked against the other from what was published. [1][2]

  • 02

    Whether psilocybin protects human nerves at all.

    Every result in the paper is in mice, and the study ran eight months. The first human trials begin this month. [1][2]

  • 03

    Whether starting the new breast cancer pill at the blood test rather than at visible growth helps anyone.

    The US drug regulator granted accelerated approval and said itself that this is not yet confirmed, ordering further studies. [4]

  • 04

    How much of the published RNA-silencing literature is overstated.

    The artifact found in Nature Biotechnology inflated the measured effect across every CRISPR system tested, and nobody has gone back through the existing papers. [19]

  • 05

    Why a quarter of the mutations that kill phi X kill it.

    It is the most studied genome there is, every letter has now been changed, and the team still cannot explain those cases. AI tools built for exactly this did badly. [18]

  • 06

    Whether the Pentagon agreement moves money out of infectious disease research.

    Members of Congress say it could; the agreement obtained by the New York Times covers chemical and biological defence work, and no figures have been published. [13]

  • 07

    Whether the genetic approach behind the failed Angelman drug can work at all.

    Ultragenyx's late-stage trial missed, and researchers not involved say one failure does not settle the mechanism. Other trials are still running and have not read out. [27]

  • 08

    Why nobody has bought Abivax.

    The company has strong late-stage results for a bowel disease pill, takeovers across the industry are running hot, and STAT's columnist says there is no visibility into what is actually happening. [28]

  • 09

    What the United States' real measles death count is.

    The health secretary and the disease agency's leadership are disputing Pennsylvania's deaths, and the case definitions that states report against are themselves now contested. [14]

04 Hope carry this

Two doses of psilocybin before each round of chemotherapy completely protected mice from the nerve damage that hits most people given platinum drugs, and the tumours shrank just the same. A version of the compound that causes no hallucinations worked about as well.

Also true today

  • In 172 adults with acute myeloid leukaemia who were fit enough for intensive chemotherapy, two gentler drugs beat it: 14.5 months before the disease returned or the patient died, against 6.2 months. Severe infections were lower too, 28% against 41%.
  • New Zealand's medicines regulator approved MDMA on Friday for people with severe post-traumatic stress disorder. It is the second country to allow it, after Australia in 2023.
  • Drug regulators from several countries finished a joint review of a test that would cut how much horseshoe crab blood the industry needs to check medicines for contamination. The crabs are currently caught, bled and returned to the sea.
  • Omega-3 with low-dose aspirin matched antibiotics over a year in people with severe gum disease, with about 58% of each group reaching the treatment goal.

Across the beats