Biotech & Longevity · Saturday, 29 August 2026

01 Briefing what happened

A heart drug failed because most patients were already on one that works

Biotech & Longevity 1 min 40 sources

AstraZeneca and Ionis explained a 1,400-person failure on Friday. Eight in ten volunteers were already taking a rival drug, and there was little harm left to prevent.

81%

already on the older drug

in the rival trial that worked, 53% [1]

29% vs 32%

heart events or deaths

drug against dummy, over 140 weeks [1]

1,400+

people enrolled

all already on standard treatment [1]

$1.9bn

half-year sales of a rival silencer

86% of Alnylam's revenue [1]

The lead story — what happened

AstraZeneca set out on Friday why its heart drug gave no extra benefit to people already on another medicine. [2]
The trial was called CARDIO-TTRansform. It enrolled more than 1,400 people and ran for 140 weeks. [1]
29% of those on the drug had a heart event or died. On dummy injections it was 32%. [1]
That gap was too small to count as a real effect, so the trial missed its main goal. [1]
The disease is ATTR-CM. A blood protein called transthyretin folds wrong and stiffens the heart muscle. [1]
Two kinds of drug already exist for it. Stabilisers hold the protein in its correct shape. [1]
Silencers, including this one, stop the body making the faulty protein. Both aim at the same protein. [1]
81% of this trial's volunteers were already on a stabiliser. In the rival trial that succeeded, it was 53%. [1]
The investigators told the New England Journal of Medicine that stabiliser use kept rising while the study ran. [1]
Jefferies analyst Michael Leuchten wrote that adding a silencer on top may simply help less than everyone assumed. [1]
Analysts said the result strengthens the case that swallowed stabilisers may beat injected silencers. [2]
Alnylam sells a silencer of its own. Its shares are down about 30% since the failure was announced in July. [1]
That one drug earned Alnylam $1.9bn in the first half of 2026, which is 86% of the company's revenue. [1]
Stifel's Paul Matteis wrote it is now hard to see a path to approval for eplontersen in this disease. [1]
A Cedars-Sinai heart specialist, Michelle Kittleson, wrote up for NEJM what the negative result changes in the clinic. [3]

What is pushing on this

Room left to improve Easing↓

eight in ten patients were already treated [1]

Doubt about the whole approach Building↑

analysts now question silencers added on top [1][2]

Money riding on the answer High→

one silencer alone earns $1.9bn in half a year [1]

Standard care moving Building↑

stabiliser use rose during the 140-week trial [1]

Who is involved

AstraZeneca and Ionis — ran the failed trial and published the post-mortem Alnylam — sells the silencer whose earlier win is now being questioned Pfizer and BridgeBio — make the stabilisers most of the volunteers were already taking Michelle Kittleson, Cedars-Sinai — wrote up for NEJM what a negative trial changes in the clinic

How it unfolded

Two years ago a rival silencer presents winning data; it is approved the following year [1]
July AstraZeneca and Ionis say the trial missed, with no detail [1]
Friday the full data appear in NEJM and at the cardiology congress [1]
Friday Alnylam shares fall as much as 5% in morning trading [1]
Next Alnylam may redesign its next trial to enrol fewer treated patients [1]

Where this points

Watch who Alnylam lets into its next-generation trial. One analyst has already suggested enrolling fewer treated patients, which would raise its odds and concede the point this failure just made. [1]

The rest of the day

38 more stories on this beat.

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

02
A one-time gene edit halved cholesterol

Fifteen patients got a single gene-editing treatment for high cholesterol. Those on the highest dose saw levels fall by half, and they were still down a year later. [4]

Why it matters — Gene editing has been for rare fatal diseases. This one goes at a common condition that daily pills already treat, and about half the people who start those pills stop. [4]
03
Pennsylvania measles cases reach 460

The state's outbreak grew to 460 cases by Friday. [6] Two unvaccinated residents of Lancaster County died last week. [8] They were the first US measles deaths of 2026 and the state's first in 35 years. [39]

Why it matters — Before the vaccine arrived in 1963 the US had 3 to 4 million cases a year and 400 to 500 deaths. [39]
04
Vaccine comments draw criticism

Robert F. Kennedy Jr.'s comments on vaccines drew criticism while the outbreak was still growing. [7] Pennsylvania's governor Josh Shapiro had called the deaths completely preventable and named the MMR shot as the best protection. [40]

Why it matters — In Lancaster County, where the deaths happened, 85% of kindergarten-age children have the shot, below the 95% the state says it needs. [40]
05
A blood cancer gets a drug instead of bloodletting

The FDA approved rusfertide, sold as Mimrylo, for polycythemia vera, a slow blood cancer affecting about 90,000 Americans. Patients have historically relied on frequent bloodletting. Takeda projects $1bn to $2bn in peak sales. [5]

Why it matters — It is the rare case where the new treatment replaces a procedure rather than another drug.
06
First dermatomyositis pill priced

Roivant and Priovant set a list price of $35,000 for a 30-day supply of Lisraya. It was approved the day before for dermatomyositis, a disease of painful skin rashes and weakening muscles. [36]

Why it matters — It is the first targeted pill for the condition, and the price arrived one day after the approval.
07
Moderna banks $2.6bn after its win

Moderna priced an enlarged convertible note offering at $2.6bn, having proposed $2bn a week earlier. It follows the phase 3 success of its cancer vaccine with Merck. [9][37]

Why it matters — The company added $45bn in market value the day those results landed, then borrowed against the mood. [9]
08
A two-drug HIV tablet cleared

The FDA approved Gilead's Bixlenvo, one tablet combining bictegravir and lenacapavir, for people whose HIV is already suppressed, including those on complicated drug schedules. [9]

Why it matters — Fewer pills is often the difference between a treatment that works and one a person keeps up with.
09
A biotech's data leaked early

Generate Biomedicines' shares fell by double digits after three embargoed posters on its asthma and lung drug golukibart were made public before a European conference. [9]

Why it matters — The findings had not changed. Only the order in which people saw them had.
10
Gates money for preeclampsia

ProFound Therapeutics secured a $35m commitment from the Gates Foundation, starting with $20m, to hunt for drug targets in preeclampsia and eclampsia using artificial intelligence. [9]

Why it matters — Both conditions are described by the foundation's own AI chief as underdiagnosed and undertreated. [9]
11
Half a billion for the world's disease ledger

The Gates Foundation pledged $540m over ten years to the Institute for Health Metrics and Evaluation, the largest grant in the University of Washington's history. [10]

Why it matters — It publishes the Global Burden of Disease study, which covers 375 diseases across 204 countries and guides where health money goes. [10]
12
The ledger's critics get louder too

Researchers say the institute's models are opaque and lean on data skewed to rich countries. A 2015 edition named cholera the leading cause of diarrhoea deaths in Canada, and was corrected without explanation. [10]

Why it matters — The same numbers steer national health spending, so a modelling error is not an academic one.
13
Twenty biotechs went public this year

Twenty drug developers have listed in 2026, against 11 in all of last year, with fourteen raising over $250m each. Bankers counted 80 acquisitions worth $96bn upfront in the first half. [12]

Why it matters — Buyers are picking approved or late-stage drugs, because $200bn of big-pharma revenue loses patent protection by 2030. [12]
14
The youngest companies miss it

A Massachusetts industry report found seed rounds shrinking while later rounds grow. The group named the earliest startups with the riskiest science as its biggest concern. Employment fell 3% year on year. [13][14]

Why it matters — The recovery is real and it is landing on companies that already had money.
15
China licensing reached $79bn

Licensing deals for drugs discovered in China totalled $79bn last year, against about $1bn in 2019. China now runs more early-stage trials than any other place the report measured. [13]

Why it matters — A US startup now competes for the same buyer against a finished Chinese asset.
16
Women's health funding fell hard

Venture money into women's health companies dropped from $3.2bn in 2024 to about $2bn in 2025. Among drug startups specifically it fell from $1.3bn to $610m. [15]

Why it matters — Investors moved to safer, later-stage bets, and the newest areas lost most.
17
Metsera's backers do it again

Arch Venture Partners and Population Health Partners incorporated Sentivera twelve days after Pfizer closed its $10bn purchase of their last startup. [16] It has licensed a Chinese inflammation drug for up to $1.5bn. [16][17]

Why it matters — The upfront cash was $40m. The rest depends on a drug only just cleared to enter human testing in China. [16]
18
Teva bids for a bankrupt biotech

BioXcel Therapeutics filed for bankruptcy after telling investors it could not fund operations past August. Teva was named the lead bidder, able to take substantially all its assets for $57.5m upfront. [11]

Why it matters — In the same week the sector index sat near a record high.
19
A weight-loss drug cut other costs

A real-world study followed adults over 55. Those who stayed on Eli Lilly's Zepbound had monthly costs $181 lower at six months and $607 lower at a year. Most of the saving came from fewer hospital and emergency visits. [17]

Why it matters — Lilly funded and announced it, and the saving depends entirely on people staying on the drug. [17]
20
A Chinese drug beats an immunotherapy

Monitors stopped a phase 3 trial early after Akeso's ivonescimab plus chemotherapy improved survival against an immunotherapy-chemotherapy combination in first-line biliary tract cancer. Summit's shares rose 14%. [17]

Why it matters — Akeso says it is the first time any therapy has beaten that combination in this cancer. No figures were released. [17]
21
Prime editing learns to swap whole genes

A tool called prime assembly cut out stretches of genome up to a million letters long. It then dropped in donor DNA of one to six thousand letters. Inserting a 2.9-thousand-letter donor worked in up to 57.8% of cultured cells. [18]

Why it matters — Editing has mostly changed single letters. Replacing a whole gene is a different scale of repair.
22
The CAR's design decides the escape

About half of patients relapse after CD19 CAR-T therapy. Researchers found the two approved products differ: repeated exposure to one design drove tumour cells to lose the CD19 target, while the other did not. [19]

Why it matters — The escape route was written into the engineering, not into the cancer.
23
A blood mutation predicts a bad CAR-T

Among 104 CAR-T recipients, having CHIP overall predicted nothing. But patients whose pre-existing clones carried TP53 mutations had lower platelets and haemoglobin and worse outcomes. [20]

Why it matters — The useful signal was invisible until the group was split by which mutation it carried.
24
Helper T cells made to order

Boston University researchers found that switching off Notch signalling late in development, while damping another signal, let stem cells mature into CD4 helper T cells at scale. [21]

Why it matters — CAR-T is built from each patient's own cells, which is why it is so expensive. Off-the-shelf cells are the way out.
25
An HIV drug improved ageing marks in mice

Old mice accumulate killer CD4 T cells in the marrow that skew blood production. Blocking their signal with maraviroc, an approved HIV drug, normalised the effect and improved several ageing measures. [22]

Why it matters — It worked in mice. Most things that work in mice never work in people.
26
Two drugs reach a hard head-and-neck cancer

In 31 patients with nasopharyngeal cancer that had already resisted platinum chemotherapy and immunotherapy, an antibody-drug pairing shrank tumours in 71%, with responses lasting a median 14 months. [23]

Why it matters — Four in ten had a serious side effect, and 31 patients is a small early trial. [23]
27
Immunotherapy before colon surgery

Twenty-four patients got two immune drugs before their colon cancer operation. In the harder-to-treat group, 41% had a major reduction in tumour tissue by the time of surgery, and no surgery was delayed. [24]

Why it matters — Immunotherapy has mostly failed in this form of colon cancer.
28
AI structure tools tested on HIV

Researchers benchmarked five structure-prediction models against experimental structures published after each model's training cut-off. AlphaFold-based models did best overall, but accuracy dropped sharply on the largest enzyme. [25]

Why it matters — Testing only on structures released after the training date is what stops a model scoring on answers it has already seen.
29
Naming a germ by its weight

Machine-learning models trained on 255 mass readings told viruses from bacteria and named individual species. They then held up against an outside database of dangerous bacteria held by the Robert Koch Institute. [26]

Why it matters — Speed of identification is what decides which antibiotic a patient gets on day one.
30
A program for stubborn drug targets

A pipeline pairing a construct-screening program with a designed marker protein solved structures of receptors that had resisted the usual trial and error, including one bound to the drug tolvaptan. [27]

Why it matters — This receptor family is behind a large share of all drugs, and its resting shape has been the hard one to see.
31
Guessing what an untested drug does

A framework called MAP joined 14 public databases into one map of 187,089 drugs, 22,924 genes and 694,246 relationships. It then predicted how cells react to compounds it had never seen, beating the best existing models by about 12%. [28]

Why it matters — Only a sliver of possible compounds has ever been tested in cells.
32
Drug development as a feedback loop

A paper in Nature Biomedical Engineering looks at the low odds of a drug surviving from first human test to approval. The authors argue the shortage is not of data. It is of turning data into decisions about what to stop. [29]

Why it matters — It proposes running development as a loop that updates, rather than a pipeline that proceeds.
33
Waiting for a drug he can see

Marc Powell, 60, has the SOD1 form of motor neurone disease, which affects fewer than 100 people in the UK. Biogen supplies tofersen free to some centres, but hospitals must fund giving it, so places are few. [30]

Why it matters — The NHS has not commissioned the drug. The barrier is the cost of administering a free medicine. [30]
34
An ultra-rare drug nears filing

Spruce Biosciences said two FDA meetings supported its plan to file in the fourth quarter for an enzyme replacement therapy for Sanfilippo syndrome type B. The filing had slipped from the first quarter over manufacturing data. [31]

Why it matters — Analysts had called the clinical data solid and the manufacturing risk hard to quantify. [31]
35
Saudi Arabia builds the factory

Saudi officials writing on their national biotech strategy argue a country can sequence its whole population and still treat nobody. Gene therapies and engineered cells have to be made to sterile standards, at scale. [32]

Why it matters — It is the same constraint that put a US cell-therapy factory maker into layoffs this week.
36
Conflict widens the Ebola damage

Doctors writing in Nature Medicine say May's Ebola emergency sits inside a wider collapse in eastern Congo. More than 7 million people there are displaced, and humanitarian funding has been cut. [33]

Why it matters — Surveillance and medicine buying were among the first functions to stop, which affects malaria, TB and HIV. [33]
37
Telling volunteers what was found

The NIH asked for public comment on a draft policy making it standard practice to share summary study results with the people who took part in the research. [34]

Why it matters — Trial volunteers currently have no general right to be told what their own study concluded.
38
A plan for bowel cancer in the young

A Nature Reviews Cancer roadmap says bowel cancer under 50 is rising worldwide. Obesity, diet and inactivity are linked to it. When and how those risks act is still unknown. [35]

Why it matters — Prevention advice cannot be written until someone knows which exposure matters and at what age.
39
Sequencing shares rose on someone else's win

Shares in several DNA sequencing companies climbed alongside Moderna and Merck after what the two companies called the first positive phase 3 result for a personalised cancer vaccine. [38]

Why it matters — Each dose is built from that patient's own tumour, so the vaccine cannot exist without sequencing. [38]

02 Lesson why it matters

Two things that each work, adding up to nothing

A drug is never scored on its own. It is scored on the harm left over after whatever people already take.

The twist

Two treatments can each work well and still add up to nothing, because they are both draining the same pool of preventable harm - and the first one got there.

How it works

A trial does not weigh a drug on its own
It weighs the drug against whatever patients already get
So the score is the harm still left to prevent
A good earlier treatment has already removed most of it
The newcomer then measures near zero
And the reading says nothing about its strength

Where you've seen this

Flood defences

a second wall on a bend the first wall already holds prevents no flood

Faster computers

a quicker processor shows no gain on a machine that is waiting for the disk

Road building

a new bypass saves no minutes on a route that was never the jam

Charity work

a second food van on a street already served feeds nobody new

The catch

This does not show the drug is weak. It shows it adds little to people already treated - which is a different question from whether it helps someone with nothing.

And the whole of it

Arrive second to a problem someone competent already handled, and you will look like you did nothing. Arrive first, and part of your credit was the size of the problem, not the size of your work. Most of us cannot tell which of the two we are, because whatever was drained before we got here left no mark.

03 Lab your turn

Build the Trial

Enrol patients who are already treated and watch the same drug measure as nothing.

04 Truth what's really going on

What is really going on

A result that reads as a failed drug is mostly a measurement of how well the older drug already works.

Why it works on us — A single word does the work here. Calling a result a failure points at the drug, when the number actually describes the gap between two treatments, and a gap has two sides.

Who gains

Pfizer and BridgeBio — They sell the swallowed stabilisers. Analysts read this failure as evidence that stabilisers beat injected silencers, and the case was made by a rival's trial. [2]
Whoever sells the first treatment in any disease — Being first means being tested against nothing. Everyone after is tested against you, on the harm you left behind. That is our reading of the pattern, not a claim in the reporting.
Investors who bought Alnylam after the July warning — The shares are down about 30% since the July announcement and fell as much as 5% more on Friday, so most of the drop came before the detail did. [1]
Established biotech companies — Buyers want approved or late-stage drugs, and $200bn of big-pharma revenue loses patent cover by 2030, so the bidding is for finished work. [12]

Who pays

Alnylam shareholders — One drug is 86% of the company's revenue, and a rival's failed trial has put a question mark over how much it adds. [1]
The newest biotech startups — Seed rounds are shrinking while later rounds grow, and one industry group named the earliest, riskiest science as its biggest worry. [13][14]
Women's health companies — Venture money fell from $3.2bn to about $2bn in a year, and among drug startups it more than halved. [15]
Marc Powell and other SOD1 patients in England — Biogen gives the drug away, but hospitals must fund the cost of administering it, so places are limited. [30]
Unvaccinated households in Lancaster County — Two people died and the outbreak has reached 460 cases, in a state that had not lost anyone to measles in 35 years. [6][39]

What nobody knows yet

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

01

Whether more people on the drug died than on the dummy injection.

One report gives 29% of the drug group against 32% on placebo, and in the same piece says more events occurred in the drug group. Both cannot be right. [1]
02

Whether the drug helps someone who is on nothing else.

The trial only enrolled people already on standard care, so it never asked. Eighty-one per cent were on a stabiliser. [1]
03

How well the rival silencer really works.

Its winning trial had 53% of patients on a stabiliser against 81% here. [1] The two drugs have never been tested against each other.
04

What the one-time cholesterol edit does after a year.

Fifteen patients have been followed for twelve months. [4] Every claim about permanence beyond that is a claim about time nobody has waited through yet.
05

How much Takeda's new blood cancer drug will cost.

The company said it would price fairly and declined to give the figure at launch. [5]
06

Why the Pennsylvania measles outbreak keeps growing.

Cases reached 460 by Friday and officials have released no details about the two people who died. [6][8]
07

How large the survival gain was in Akeso's stopped cancer trial.

Monitors halted it early and the company called the result meaningful without publishing any numbers. [17]
08

Whether the global disease ledger's estimates can be checked.

Researchers say the models are opaque and neither easily verifiable nor replicable, and a past error was corrected without an explanation. [10]

05 Hope carry this

A trial that fails because the people in it were already being treated well is a strange kind of good news. It means the ground moved under the question while nobody was watching.