Day Lila

Biotech & Longevity · Friday, 25 September 2026

01 Briefing what happened

Merck's new eye drug did no worse than Lucentis in a year-long trial. More patients had bleeding inside the eye.

Biotech & Longevity 16 sources

Remigromig, a drug for eye swelling caused by diabetes, works differently from today's drugs. It matched Lucentis on vision after a year, but had more bleeding and more patients stopping. Merck says it is aimed at the 30% to 40% of patients today's drugs fail.

30% to 40%

of patients do not respond to today's eye drugs, or stop responding

Merck's research chief gave the range; remigromig is aimed at them [1]

52 weeks

until vision was compared on an eye chart

both remigromig doses were judged no worse than Lucentis [1]

$3bn

the most Merck can pay for EyeBio, which made remigromig

$1.3bn was paid upfront in 2024, and up to $1.7bn more depends on goals being met [1]

The lead story — what happened

  • Merck said its experimental eye drug remigromig did no worse than Lucentis, an established treatment, in Brunello, a trial that ran for a year. [1]
  • The patients had diabetic macular edema: swelling at the back of the eye in people with diabetes, which blurs the centre of their vision. [1]
  • Lucentis, sold by Roche's Genentech in the US, blocks VEGF, a signal that makes blood vessels in the eye grow and leak. [1]
  • Remigromig works another way. It copies norrin, a natural molecule that helps keep the wall between the blood and the retina sealed. [1]
  • The trial's main question was how well patients read an eye chart after 52 weeks. Both doses of remigromig were judged no worse than Lucentis. [1]
  • More patients on remigromig had new fragile blood vessels, bleeding into the gel inside the eye, or stopped because of side effects. [1]
  • The report gives no numbers for those side effects, and Merck says it is running more analyses. [1]
  • In an earlier, smaller trial run by EyeBio, the company that first made the drug, no side effects were linked to it. [1]
  • Merck's research chief, Dean Li, said 30% to 40% of patients do not respond to today's VEGF drugs, or stop responding. [1]
  • He said Merck expects eye doctors to mix VEGF drugs and non-VEGF drugs like remigromig. [1]
  • A Wells Fargo analyst asked Li whether Merck could beat Lucentis, and raised the argument that Merck had picked the weakest rival. Lucentis competes with Regeneron's Eylea and Roche's Vabysmo. [1]
  • Merck will present the data at the American Academy of Ophthalmology meeting next month and discuss them with regulators. [1]
Two drugs aimed at the same eye swelling by two different routes. The trial asked whether the new route was at least as good.

Who is involved

  • Merck

    one of the world's biggest drugmakers; it bought remigromig and ran the Brunello trial

  • Roche and Genentech

    the Swiss drugmaker and its US arm; they sell Lucentis and a rival drug, Vabysmo

  • EyeBio

    the small company that first made remigromig; Merck bought it in 2024

  • Dean Li

    head of research at Merck; he says the drug is for the patients today's drugs fail

How it unfolded

  1. Earlier EyeBio's small early trial links no side effects to the drug [1]
  2. 2024 Merck buys EyeBio for $1.3bn upfront [1]
  3. This week Brunello: remigromig no worse than Lucentis, with more eye bleeding [1]
  4. Next month full data due at the American Academy of Ophthalmology meeting [1]
  5. March a second trial against Lucentis is due to finish [1]

Where this points

Watch the full numbers at next month's eye-doctors' meeting, above all how many patients had bleeding inside the eye, and whether a second trial against Lucentis matches Brunello in March. [1]

What is pushing on the whole day

The bar and the word are our reading of how hard each one is pushing today. The arrow is where it is heading. The evidence is in the stories below.

Drugs for the patients others miss Building↑

Merck says 30% to 40% of eye patients get too little from today's drugs, and aimed remigromig at them. [1] Kyverna's CEO said every drug now used for stiff person syndrome is borrowed from other diseases. [2] Acadia designed a successor to its older psychosis drug to carry less heart-rhythm risk. [3]

Side effects deciding the result High↑

Remigromig matched Lucentis on vision but had higher rates of bleeding in the eye. [1] Amberstone designed its cancer drugs to work mainly in a tumour's acid, to spare healthy tissue. [6] Adding evorpacept in stomach cancer brought more blood-count side effects. [4]

Small trials, big claims Building→

A Huntington's eating study had 20 people and no comparison group. [10] A spinal-injury trial of magnetic brain stimulation had 29. [13] Kyverna's one-year result compares 26 patients with how they were before treatment. [2]

Money for early cancer drugs Building↑

Novartis agreed to pay up to $900m for a radioactive cancer drug not yet tested in people. [5] Henlius, a Shanghai drugmaker, agreed to pay up to $440m for drug designs Amberstone has yet to make. [6] City Therapeutics filed to sell shares on Nasdaq. [8]

The rest of the day

15 more stories on this beat.

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

  1. 02

    A one-time cell treatment keeps patients walking

    Kyverna Therapeutics reported one-year results on 24 September for miv-cel, a CAR-T treatment. [2] CAR-T means taking a patient's own immune cells, changing them to hunt a target, and giving them back. [2] In 26 people with stiff person syndrome, a rare disease of rigid muscles and painful spasms, walking speed improved by a median 49%. [2] Eight of the 12 who needed a walking aid before still did not. [2] In a second disease, myasthenia gravis, all five patients followed for a year kept their improvement. [2]

    8 of 12

    patients who needed a walking aid before treatment and no longer did a year later

    Twelve patients used a walking aid before the one-time treatment. Eight still did not need one a year on.

    Why it matters — If the US regulator approves it, it would be the first CAR-T for a disease where the immune system attacks the body. [2] Rivals Novartis and Bristol Myers Squibb paused similar trials in August after dangerous inflammation, including three patient deaths in Novartis's trials. [2]

  2. 03

    An Alzheimer's psychosis drug just misses

    Acadia Pharmaceuticals said remlifanserin missed its main goal in the first part of a trial in people with Alzheimer's disease who have hallucinations and delusions. [3] On the higher dose, a symptom score fell 12.6 points by week 6, against 10.4 on a dummy drug. [3] The result fell just short of the line trials use to rule out chance. [3] The drug was built to avoid a heart-rhythm risk of Acadia's older drug Nuplazid, and none was seen. [3]

    How far the symptom score fell by week 6. The dummy-drug group improved almost as much.

    Why it matters — Acadia will still run the final-stage part, with only the higher dose, and its shares opened 11% lower. [3] Analysts noted that drugs borrowed from other conditions for these patients can cause lasting movement problems in some of them. [3]

  3. 04

    A stomach-cancer add-on beats one test, not two

    Nature Medicine published a trial of evorpacept in 127 people whose advanced stomach cancer is driven by a protein called HER2 and had already been treated. [4] Evorpacept blocks CD47, a signal cancer cells use to stop immune cells from eating them. [4] Added to three standard drugs, it shrank tumours in 40.3% of patients, against 26.6% without it. [4] Where a fresh biopsy still showed HER2, the figures were 54.8% and 23.1%. [4]

    Share of patients whose tumours shrank, across the whole trial.

    Why it matters — It beat the comparison group by the margin set in advance, but missed a second test against a 30% benchmark from past trials. [4] Blood-count side effects were more common, though overall safety was similar. [4]

  4. 05

    Novartis licenses a Chinese radioactive drug

    Novartis, the Swiss drugmaker that leads the market in radioactive cancer drugs, agreed this week to pay BoomRay Therapeutics of Suzhou, China, up to $900 million. [5] The money covers an upfront payment plus milestones and royalties for a drug not yet tested in people. [5] These drugs carry a radioactive atom to a tumour. Novartis sells two: Pluvicto for prostate cancer and Lutathera for rarer hormone-cell tumours. [5]

    Why it matters — The deal came a day after Telix, an Australian company, agreed to buy ITM, whose drug would compete with Lutathera. [5] Novartis had dropped a mid-stage radioactive drug in July after disappointing results. [5]

  5. 06

    Henlius pays for designs from a US biotech

    Shanghai Henlius, a Chinese drugmaker that built its business on cheaper copies of biologic drugs, agreed a deal worth up to $440 million with Amberstone, a US biotech. [6] Amberstone will design T-cell engagers for up to two targets Henlius chooses. [6] A T-cell engager is an antibody that ties the body's killer T cells to a cancer cell. [6] Most approved ones treat blood cancers. [6]

    Why it matters — Their known danger is an immune overreaction called cytokine release syndrome. [6] Amberstone says its designs switch on in the acid around solid tumours and far less elsewhere, so far only in lab tests; its own first drug should reach people next year. [6]

  6. 07

    Price deals may shrink Medicare's savings

    The Trump administration has signed drug-price deals with 26 companies, tying some prices for Medicaid, the US health plan for people on low incomes, to what other rich countries pay. [7] Two separate plans for Medicare, the US government's health plan for older people, would also peg US prices to other countries. [7] A Lancet study estimated those plans would save about $11.6 billion a year at first. [7] Some drugmakers say their deals exempt them, and the study's lead author estimates exemptions could cut the savings by nearly 80%. [7]

    Why it matters — The Medicare plans, called GLOBE and GUARD, are due to start on 1 October and 1 January 2027, but are not final. [7] The US government has not said which companies are exempt, and the deals' details are confidential. [7]

  7. 08

    An RNA drug maker files to list

    City Therapeutics, founded in 2023 in Cambridge, Massachusetts, filed on 24 September to sell shares on the Nasdaq stock market. [8] It makes drugs based on RNA, the molecule that carries a gene's instructions, for rare diseases. [8] Its lead drug, CITY-FXI, is in an early trial against thrombosis, harmful clots inside blood vessels. [8] It has raised about $238.8 million from investors so far. [8]

    Why it matters — ADARx, another RNA drug company, was expected to list on Friday. [8] Reuters reported that autumn listings have struggled as rising interest rates on government bonds and worries about AI shares weigh on investors. [8]

  8. 09

    A heart-device maker cuts most of its staff

    Adagio Medical, which makes a device that freezes heart tissue to stop dangerous fast heart rhythms, let go 25 of its 43 full-time staff. [9] Its board is exploring a sale or merger. [9] Its shares closed at 23 cents on Wednesday, down 53%. [9] It had about $7.7 million in cash in June, and Nasdaq warned it in August about its finances. [9]

    Why it matters — The device uses liquid nitrogen at minus 196 degrees Celsius, and Adagio is still asking the US regulator to approve it. [9] It sent in its main trial results in May. [9]

  9. 10

    Eating within a few hours, tried in Huntington's

    Researchers at Oregon Health & Science University asked 20 people with early Huntington's disease to eat only within a six-to-eight-hour window each day for 12 weeks. [10] Huntington's is an inherited brain disease with no approved treatment that slows it. [10] A disease-severity score improved by 0.5 points on average; it usually worsens about one point a year. [10] A blood marker of nerve damage fell 13%. [10]

    Why it matters — The study was small and had no comparison group, so it cannot show the eating pattern slows the disease. [10] The team is seeking money for a proper trial against normal eating. [10]

  10. 11

    Long COVID linked to lost dopamine nerve endings

    Scientists at Canada's Centre for Addiction and Mental Health scanned the brains of people with long COVID and of healthy people. [11] Dopamine is a brain chemical involved in motivation and movement. [11] People with long COVID had much less of a marker of dopamine nerve endings in the striatum, a group of brain regions. [11] Losses in different areas matched different symptoms. [11]

    Why it matters — Long COVID is estimated to affect about 5% of people worldwide and has no proven treatment. [11] The team plans a trial of drugs that boost dopamine in the coming months. [11]

  11. 12

    Gut bacteria that stop maturing, and diabetes

    A study in Nature Metabolism followed 887 children at high genetic risk of type 1 diabetes for up to six years. [12] Researchers read the DNA of gut microbes in 12,151 samples. [12] Children whose gut microbes stopped maturing early had about three times the risk of type 1 diabetes. [12] Type 1 diabetes happens when the immune system destroys the cells that make insulin. [12]

    Why it matters — It shows a link, not a cause. [12] It suggests early microbes and a child's own genes act together in who gets the disease. [12]

  12. 13

    Magnetic pulses with therapy after neck injury

    A trial published on 22 September tested magnetic pulses to the brain, alongside occupational therapy, in 29 people with spinal cord injuries in the neck. [13] Half got real pulses and half got a fake version, three times a week for five weeks. [13] The real group gained more arm and hand strength, feeling and independence in daily life. [13]

    Why it matters — The trial is small, with 14 and 15 patients in its two groups. [13] Scores for pain, muscle stiffness and low mood rose in both groups. [13]

  13. 14

    One blood spot, 1,200 DNA checks for newborns

    Researchers described a test that checks up to 1,200 DNA targets from a single punch of a newborn's dried blood spot. [14] It was tried on samples from 2,069 babies in the Philippines. [14] It found gene changes behind conditions such as spinal muscular atrophy and fragile X, and signs of a severe immune disorder, in one run. [14] Results take 36 to 40 hours. [14]

    Why it matters — The authors built it for countries where DNA sequencing is too costly, or too contested, for public newborn screening. [14]

  14. 15

    How muscle channels open together

    Scientists at Berlin's Max Delbrueck Center imaged a muscle calcium channel, RyR1, at six stages of opening inside its natural membrane, using rabbit muscle. [15] The channel releases the calcium that makes a muscle contract. [15] Neighbouring channels touch and help each other open, which one scientist compared to cogs in a clock. [15]

    Why it matters — Many gene changes behind malignant hyperthermia, a dangerous reaction to some anaesthetics, sit exactly where the channels touch. [15] That contact point could become a target for new drugs. [15]

  15. 16

    A Chinese drug factory passes a European inspection

    Altruist Biologics, a Chinese company that makes drugs for other firms, said its Suzhou factory passed an inspection by the European Medicines Agency with no critical findings. [16] It is the first time the site has passed. [16] The inspection supports a client's application to sell a biologic drug in the European Union. [16] Altruist says it holds about 20% of China's capacity of this kind. [16]

    Why it matters — European approval of the client's drug depended on this inspection. [16] The US regulator recently rejected ITM's cancer drug on manufacturing grounds, before Telix agreed to buy ITM. [5]

02 Lesson why it matters

The next drug is built for the patients the last one missed

A first drug for a disease helps most people but not all, so the next drug is designed for the ones it left behind.

The twist

A new drug does not have to beat the old one for everybody. It has to help the patients the old one did not help.

The picture

3 of 10

patients with diabetic eye swelling who get too little from today's drugs

Merck puts it at 30% to 40% of patients. The drawing shows the lower end: 3 people in every 10.

How it works

  1. A first drug helps most people with a disease
  2. Some people get too little from it, or cannot take its side effects
  3. Those people become the reason to build a next drug
  4. The next drug has to work a different way
  5. It is tested against the old drug and must at least match it
  6. Doctors then use both, picking for each patient

The same force, elsewhere today

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

  • Acadia's psychosis drug

    Nuplazid carries a heart-rhythm risk, so Acadia built remlifanserin to hit the same target without it.

  • Kyverna's cell treatment

    Stiff person syndrome is treated with drugs borrowed from other diseases, and Kyverna built a one-time treatment for the patients those drugs do not stop.

  • Evorpacept in stomach cancer

    It is added for patients whose cancer grew despite HER2 drugs, and it works through a different part of the immune system.

  • Amberstone's cancer drugs

    Today's T-cell engagers can harm healthy tissue, so Amberstone designed ones that work mainly in a tumour's acid.

Where you've seen this

Contact lenses

made for people who could not get on with glasses, not to replace glasses for everyone

Night buses

run for the riders the daytime timetable leaves without a way home

Gluten-free bread

sold to the people ordinary bread makes ill

The catch

This only pays if doctors can tell early who the old drug will fail. If they cannot, the new drug competes for everyone and must beat the old one outright.

And the whole of it

Every disease has patients the standard treatment does not help, and usually nobody knows who they are until the treatment has failed. Any of us could turn out to be one of them.

03 Truth what's really going on

What is really going on

Several of this week's new drugs are aimed at patients whom existing treatments fail. Merck's eye drug only had to show it was no worse than Lucentis, and Kyverna's cell treatment was measured against each patient's own starting point, with no comparison group. [1][2]

Why it works on us — The trade press called Merck's result a pivotal victory, and a victory sounds like better than Lucentis when the result was no worse than Lucentis. [1]

Who gains

  • Merck — A second route into the market for diabetic eye swelling, part of its effort to grow as patent protection on its cancer drug Keytruda runs out. [1]
  • Kyverna — Its rivals paused their trials, while Kyverna reports no severe reactions in over 100 patients treated so far. [2]
  • Drugmakers that signed price deals — Some say their deals exempt them from Medicare's planned price limits, which were meant to cut what Medicare pays. [7]
  • BoomRay and Amberstone — Small companies are being paid by bigger ones, up to $900m and $440m, for drugs not yet tested in people. [5][6]
  • Altruist Biologics — Passing a European inspection lets it make drugs its clients can sell in the European Union. [16]

Who pays

  • Patients in the Brunello trial — More of those given remigromig had bleeding inside the eye or stopped because of side effects. [1]
  • Medicare and US taxpayers — Exemptions could cut the planned savings on drug prices by nearly 80%, on one researcher's estimate. [7]
  • Adagio Medical's staff — 25 of 43 full-time employees lost their jobs as the company looks for a buyer. [9]
  • Alzheimer's patients with hallucinations — They wait longer for a drug made for them, while drugs borrowed from other conditions can cause lasting movement problems in some. [3]
  • Acadia's shareholders — The shares opened 11% lower after the near miss. [3]

What nobody knows yet

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

  • 01

    How many patients on remigromig had bleeding inside the eye.

    The report says the rates were higher than on Lucentis but gives no numbers, and Merck is still analysing them. [1]

  • 02

    Whether remigromig helps the patients Lucentis fails.

    Merck's research chief named those patients as the target. The reported result compares the two drugs across all the patients in the trial. [1]

  • 03

    How much of Kyverna's result is the treatment itself.

    The 26 patients were measured against their own starting point, not against a group given something else. [2]

  • 04

    Why the rival cell treatments caused dangerous inflammation.

    Analysts pointed to the fast way Novartis and Bristol Myers Squibb make their cells, but that was their guess. [2]

  • 05

    Whether Acadia's final-stage trial will clear the line its first part missed.

    Acadia is considering enrolling patients with somewhat worse symptoms, and it tested that idea only by re-reading data it already had. [3]

  • 06

    Which drugmakers are exempt from the new Medicare price plans.

    The US government has not said, and the price deals themselves are confidential. [7]

  • 07

    Whether eating within a few hours slows Huntington's disease.

    The study had 20 people and no comparison group, and a larger trial is not yet funded. [10]

  • 08

    Whether dopamine drugs help people with long COVID.

    The scans show a loss, and the trial to test the drugs has not started. [11]

  • 09

    What Novartis and Henlius are really paying.

    Both deals give only a top figure, with no split between money paid now and money that depends on success. [5][6]

04 Hope carry this

Eight of the 12 people with stiff person syndrome who needed a walking aid before one infusion of Kyverna's cell treatment were walking without it a year later. All five people with the muscle disease myasthenia gravis followed for a year after the same treatment kept their improvement.

Also true today

  • A new newborn test checks up to 1,200 DNA targets from a single punch of a dried blood spot, and returns results in 36 to 40 hours.
  • In people whose stomach cancer still carried the HER2 protein, tumours shrank in 54.8% of those given evorpacept with standard drugs, against 23.1% without it.
  • Twenty people with early Huntington's disease ate within a six-to-eight-hour window for 12 weeks, and a blood marker of nerve damage fell by 13% on average. In Huntington's that marker usually rises.

Across the beats