Howdy, and good morning. Friday is here, so lend me your ears before you escape into the weekend. I come bearing a few interesting developments and, as usual, science has not been sitting quietly.
There is a natural instinct, when the world becomes unfamiliar, to look back towards what we know. Policy does it constantly. New science arrives and we reach for old categories. Genetics starts telling us something about personality, and somebody asks whether it belongs in the algorithm. Autism biology starts breaking one diagnosis into different molecular pathways. CAR-T begins manufacturing itself inside the patient. Gene editing raises the possibility of replacing decades of cholesterol tablets with one intervention. Magic mushrooms return as possible medicine, carrying fifty years of regulatory baggage. We crack an “undruggable” cancer target, and immediately confront the price of success. And governments proclaim AI sovereignty, until the invoice for the infrastructure arrives.
Different stories. Same tension. Science keeps moving. Policy keeps looking for somewhere familiar to put it.
That instinct is understandable. The familiar contains things worth protecting: safety, evidence, affordability, accountability and public trust. But eventually the house becomes too small.
The job of policy is not to predict precisely where science will go. It is to become much better at recognising when science has changed the terms of the argument, and adapting accordingly. And this week, science seems particularly restless. So, before everyone heads for the door, let us see where it has taken us.
In this issue
- Your Genome Has Taken a Personality Test: A vast genetic study maps personality-associated variants—and opens an uncomfortable debate about prediction, discrimination and who gets placed inside the algorithm.
- Autism: Stop Funding the Label. Start Following the Biology: Researchers are tracing different autism-linked mutations towards shared molecular machinery, challenging policy to fund pathways rather than one administrative diagnosis.
- The Underpants Test: This year’s Ig Nobels remind research funders that transformative science does not always arrive wearing a respectable suit.
- When the Patient Becomes Both the Factory and the Readout: In-vivo CAR-T and circulating tumour DNA are turning the patient into both the place where treatment is made and the system that reveals whether it works.
- The Last Cholesterol Pill?: Early gene-editing results raise the possibility of replacing lifelong medication with a single intervention, and decades of follow-up.
- Magic Mushrooms: Did Regulation Get There Before the Science?: Preclinical research suggests psilocybin could protect against chemotherapy-induced nerve damage, forcing regulators to reconsider what earlier drug classifications may have buried.
- We Cracked RAS. Now Don’t Break the Investment Model: A breakthrough pancreatic cancer medicine delivers striking survival results—and a price that tests the contract between innovation, investment and access.
- Europe Wants AI Sovereignty. Just Don’t Send the Bill: Europe wants independent computing power for AI, genomics and digital health; its finance ministries must now decide whether sovereignty is worth paying for.
| This is not another health news digest. It’s a twice-weekly readout of where evidence meets power and where power must turn into action. |
Your Genome Has Taken a Personality Test
Somewhere in a ministry, somebody will eventually ask the obvious question: Can we use this? A major study has identified 1,260 genetic variants associated with the Big Five personality traits — extraversion, agreeableness, conscientiousness, neuroticism and openness — using data from up to 1.14 million people. Scientifically important. Politically complicated. [Nature]
Governments love predictors. Health ministries want to anticipate disease. Finance ministries want to anticipate costs. Insurers calculate risk. Employers assess suitability. Give institutions a variable carrying the reassuring label “genetic”, and somebody will want to put it into the model. But this is not a gene for being sociable, anxious or conscientious. Individual effects are small, personality remains strongly shaped by environment, and the danger begins when a population-level association becomes an individual administrative judgement. There is also an equity warning. Around 95% of participants had European-like genetic ancestry, with weaker predictive performance in African-like populations. If such tools eventually influence insurance, employment, healthcare or public services, uneven evidence can become uneven treatment. The algorithm may appear impartial. The underlying evidence may not be. For personalised medicine, the dividing line matters. Using genomics to identify disease risk, guide treatment or improve screening is personalisation. Using genetics to decide that somebody is less compliant, more risky or less suitable is biological classification. Prediction is not permission.
Autism: Stop Funding the Label. Start Following the Biology.
That same problem — one administrative label concealing much greater biological complexity — is appearing in autism. Researchers have mapped the proteins encoded by 100 high-confidence autism-associated genes and identified more than 1,800 protein-protein interactions, 87% of them previously unknown. Importantly, different genetic mutations appear to converge on a smaller number of shared protein complexes. Researchers have also received $46 million from the Aligning Research to Impact Autism initiative to move the work from molecular mapping towards therapeutic development through an international programme. That potentially changes the therapeutic question. [Science]; [ARIA grant announcement]
Instead of looking for a separate treatment for every mutation, researchers may eventually be able to target shared molecular machinery across groups of patients. Autism is extraordinarily heterogeneous. Some people require relatively little support; others have substantial communication, intellectual, neurological and lifelong care needs. Precision medicine does not resolve that complexity by discovering “the autism gene”. It asks a more useful question: Do particular patients share a biological pathway that can actually be targeted? There is also an important investment lesson. That bridge matters. Governments routinely fund discovery and then wonder why discoveries fail to become treatments. Translation requires its own infrastructure: molecular profiling, longitudinal data, trial networks, regulatory pathways and sustained investment. The future increasingly looks less like diagnosis → generic intervention and more like diagnosis → molecular stratification → targeted intervention. But the policy mistake would be to choose between biology and support. People need better science tomorrow and appropriate services today. We need both.
The Underpants Test
Which brings us to underwear. Would a government funding panel approve a proposal to bury hundreds of pairs of underpants in the ground? Probably not. Which is precisely why IAPM may need an Underpants Test. Scientists have used buried cotton underwear to measure soil biological activity. Other apparently eccentric research has involved kissing, splash-free urinals and mosquito mouthparts used for high-resolution 3D printing. All featured among this year’s Ig Nobel prizes. [Nature — 2026 Ig Nobel Prizes]; [University of Zurich — buried-underwear study]
The lesson is not that ridiculous ideas are automatically good. It is that important ideas do not always arrive looking important. Yet research systems increasingly demand strategic alignment, measurable outcomes, economic return and implementation plans before the work has even started. Politicians want outcomes. Civil servants want indicators. Finance ministries want numbers. Researchers quickly learn to describe uncertainty as though it were a five-year programme. That makes funding easier to administer. It may also make science safer, narrower and less surprising. So perhaps policymakers should occasionally ask: What are we rejecting today because we cannot yet explain what it will be useful for tomorrow? That does not mean abandoning evidence or accountability. It means remembering that predictable science rarely changes the world.
When the Patient Becomes Both the Factory and the Readout
Meanwhile, personalised medicine itself is becoming less predictable. In China, five patients with treatment-refractory systemic lupus erythematosus received an experimental therapy that delivered genetic instructions directly into the body, creating CAR-T cells in vivo rather than manufacturing them externally. Conventional CAR-T requires cells to be removed, transported, genetically engineered, expanded, tested and returned. If much of that process can eventually happen safely inside the patient, one of personalised medicine’s most sophisticated technologies could become considerably easier to scale. At the same time, circulating tumour DNA is emerging as a potential real-time readout of cancer treatment response. In resectable lung cancer, patients receiving perioperative nivolumab were nearly twice as likely to clear detectable ctDNA before surgery as those receiving chemotherapy plus placebo — 66% versus 38% — with clearance associated with better outcomes. [New England Journal of Medicine / PubMed]; [Nature — CheckMate 77T biomarker analysis]
Put those developments together and something fundamental is happening. The patient is becoming both the manufacturing site and the monitoring site. Health systems, however, still organise care largely around fixed treatments, fixed diagnostic moments and reimbursement of individual products. What happens when treatment becomes a continuing biological process and a blood test can tell us whether it is working, whether residual disease remains, or whether therapy should be intensified or stopped? The question ceases to be simply: Which medicine should we reimburse? It becomes: Which treatment-plus-diagnostic pathway should we fund, monitor and adapt? Medicine is becoming dynamic. Our systems remain remarkably static.
The Last Cholesterol Pill?
Cardiovascular medicine could bring that disruption to millions of people. For decades, prevention has depended on a simple but fragile instruction: Take this medicine every day. Possibly for life. Now imagine replacing that with one intervention. An early Phase Ia study used CTX310 to edit the ANGPTL3 gene in the liver. Among the four patients receiving the highest dose, mean LDL cholesterol fell by 53% and triglycerides by 48%, with the reductions still present a year later. This is nowhere near replacing statins. But it points towards a fundamentally different model of prevention. [New England Journal of Medicine]
Instead of asking, “Will this patient still be taking treatment in ten years?”, we may eventually ask, “Should we make a one-time genetic change today?” That solves one problem by creating another. A tablet can be stopped. A gene edit is intended to endure. Patient selection, consent, evidence thresholds and long-term surveillance therefore become substantially more important — particularly as genome editing moves from rare diseases towards common conditions affecting otherwise healthy populations. Then comes reimbursement. Health systems know how to buy tablets. They know rather less about paying today for an intervention whose value may accumulate over twenty or thirty years. The patient may receive treatment once. The health system may have to follow the patient for decades.
Magic Mushrooms: Did Regulation Get There Before the Science?
Sometimes the problem is not science moving somewhere new. It is science returning somewhere policy thought it had already settled. For decades, policymakers largely knew what to do with psilocybin. Control it. Now preclinical research in mice suggests psilocybin may help prevent chemotherapy-induced peripheral neuropathy while preserving chemotherapy’s anti-tumour activity. Even more intriguingly, a non-hallucinogenic compound acting through the same biological pathway produced similar protection. This is early science, not a justification for self-medication. But the policy question is fascinating. Did regulation get ahead of the science — and then make it harder for the science to catch up? [Science]; [MD Anderson Cancer Center]
The criticism is not that regulators decades ago should somehow have predicted a role in chemotherapy neuropathy. They could not. It is that one known feature of psilocybin — its psychoactive effect — came to define the molecule itself. Once a substance is regarded primarily as a dangerous recreational drug, researching it becomes administratively harder, politically less attractive and commercially uncertain. Regulators need to control misuse. But controlling misuse should not mean closing scientific inquiry. Today’s suspicious substance may contain tomorrow’s therapeutic mechanism. The failure would not have been missing a cancer treatment fifty years ago. It would be believing that once the drug had been classified, the science had been classified too.
We Cracked RAS. Now Don’t Break the Investment Model.
Then comes the uncomfortable question that eventually follows almost every breakthrough: Who pays? Daraxonrasib, the first broad RAS-targeted medicine approved in the United States for metastatic pancreatic adenocarcinoma, delivered a striking result: median overall survival of 13.2 months versus 6.7 months with chemotherapy in previously treated patients. Then came the price: a wholesale acquisition cost of $39,800 for 30 days — $477,600 a year before rebates, discounts or patient assistance. It is easy to make this simply a story about expensive medicines. That misses half the story. [US Food and Drug Administration]; [Revolution Medicines SEC filing — price]
RAS was considered effectively “undruggable” for decades. Turning that biology into a therapy required years of scientific failure, chemistry, clinical development, manufacturing and enormous amounts of capital without any guarantee that the investment would succeed. Pancreatic cancer needs more investment, not less. If returns from high-risk oncology become unattractive, capital will move somewhere easier. But neither can the policy answer be: Pay whatever is asked. The social contract around innovation has to work both ways. Investors need confidence that genuine breakthroughs will be rewarded. Health systems need confidence that rewards reflect meaningful clinical benefit. Patients need confidence that the breakthrough will actually reach them. That points towards earlier dialogue between developers, payers and HTA bodies; outcomes-based reimbursement; appropriate differential pricing; and diagnostics being planned alongside the medicine rather than added later. The conversation needs to start during development — not when the invoice arrives after approval. We have begun to crack RAS. Now policy needs to crack the investment-and-access equation.
Europe Wants AI Sovereignty. Just Don’t Send the Bill.
All of these ambitions eventually arrive at infrastructure. Europe wants to become an AI power. There is one awkward detail: somebody has to pay for it. The EU has launched a call for tenders to establish up to seven AI Gigafactories, intended to expand European computing capacity and reduce technological dependence. But Member States are simultaneously fighting over the next EU budget, with competitiveness competing against agriculture, cohesion, defence, climate and numerous national priorities. So Europe confronts another contradiction. AI sovereignty is considerably cheaper in a strategy document than in a finance ministry. [European Commission]; [Reuters — EU budget negotiations]
For health, this is not an abstract technology debate. Genomics, digital pathology, cancer imaging, drug discovery and AI-enabled screening depend on computing capacity. If wealthy countries build the infrastructure while others cannot participate, Europe risks reproducing digitally the inequalities it has spent years trying to reduce clinically. And the same contest is playing out globally. The United States, China, the Gulf states and countries across Asia are investing heavily in compute, data infrastructure, genomics and AI. Regions without comparable capacity risk becoming consumers of models trained, governed and commercialised somewhere else. Not every country needs its own gigafactory. But countries need meaningful access to the infrastructure shaping tomorrow’s medicine. Data without compute are an archive. AI without infrastructure is a presentation. Sovereignty without investment is a slogan.
The IPM Take
Stop Waiting for Science to Come Home
Taken together, these stories are not simply about better drugs, better diagnostics or cleverer algorithms. They are about categories changing. A genetic association becomes an administrative temptation. One diagnosis fragments into multiple biological pathways. The patient becomes the factory and the readout. Prevention begins to look like permanent intervention. A prohibited substance returns as a therapeutic candidate. A breakthrough creates an investment-and-access dilemma. And data become useful only when somebody invests in the infrastructure capable of turning them into knowledge.
Policy cannot predict all of this. Nor should it try. What it can do is become more adaptive. Fund curiosity as well as translation. Regulate risk without closing scientific questions. Reward investment without making successful innovation inaccessible. Build diagnostics alongside therapies. Design reimbursement around outcomes and pathways rather than isolated products. Invest in infrastructure before lack of infrastructure becomes the implementation barrier. And recognise that these are global questions. Different health systems will move at different speeds, with different resources and political traditions. But patients should not be condemned to yesterday’s medicine simply because their institutions adapted more slowly than the science.
This is ultimately the implementation challenge at the heart of personalised medicine. A breakthrough is not complete when the paper is published. It is not complete when the regulator approves the medicine. It is not even complete when somebody agrees to pay for it. It is complete when the pathway allows the right patient to benefit. The shortcut is not the pathway. But neither should yesterday’s pathway be preserved simply because it feels familiar.
Science will continue to leave the familiar behind. The job of policy is not to drag it home, nor to follow blindly. It is to build the next pathway — protecting evidence, access and public trust while allowing innovation to move. Science has already left home. Policy should stop asking when it is coming back.
September Watch
Registrations also remain open for upcoming IPM Alliance events in New York on 24 September and Dublin on 2 October. Visit the IPM Alliance website for programme updates.
| 8 September | New York — UNGA 81 opens |
| The political machinery of the UN restarts, placing global health, financing and international cooperation back on the diplomatic stage. (United Nations General Assembly) |
| 9 September | Online — WHO puts safety inside NCD policy |
| The global World Patient Safety Day webinar will examine unsafe care in noncommunicable diseases and launch the 2026 campaign. (WHO) |
| 10–11 September | London — Precision oncology goes molecular |
| ESMO’s MAP Congress brings biomarkers, molecular diagnostics and targeted-treatment development into the same room. (ESMO) |
| 10 September | Global — World Suicide Prevention Day |
| The final year of the “Changing the Narrative on Suicide” campaign keeps prevention, stigma and mental-health investment on the public-health agenda. (WHO) |
| 11 September | Online — EMA meets the affordable-medicines sector |
| The regulator’s bilateral meeting with Affordable Medicines Europe offers a useful signal on access, supply and stakeholder priorities. (EMA) |
| 12–15 September | Seoul — Lung cancer’s biggest meeting begins |
| WCLC 2026 opens with more than 2,300 abstracts and major data expected across biomarkers, targeted therapies and immuno-oncology. (IASLC WCLC 2026) |
| 13 September | Global — Sepsis demands investment |
| World Sepsis Day shifts from awareness towards financing prevention, early detection, workforce capacity and stronger health systems. (World Sepsis Day 2026) |
Follow the last-mile conversation!
IPM Alliance tracks the decisions, bottlenecks and implementation gaps shaping personalised medicine access. Follow us for briefings, events, policy signals and practical insights from across regions and disease areas.
We want to hear from you
When a patient becomes eligible for personalised care, where does the system most often lose them: identification, testing, referral, reimbursement, delivery capacity or equity?
Know someone who should be reading this?
IPM Brief is built for people working where science, policy and patient access collide. If one colleague would find this useful, forward it to them and help bring them into the conversation.
International Alliance for Personalised Medicine
Avenue de l’Armée / Legerlaan 10, 1041 Brussels, Belgium
