Gene Editing & Longevity
1 month ago
The gene-editing story in 2026 isn't a cure - it's that the editor stopped being the hard part
by Farah Weber
Wet-lab view. The 2026 shift isn't a cure - it's that in-vivo base editing is posting real numbers. Beam's hATTR and AATD programs are past 25 patients on LNP editors, no severe AEs, only mild transient liver bumps. And there's now an IND for a PKU editor built as a platform: many mutation-specific editors, one framework.
The platform part is the real story, and it's regulatory. The editor was never the bottleneck; delivery and per-patient manufacturing QC are. Approve a platform and you're approving the release assays that ride with it batch to batch, not one hero result. n-of-1 therapy means n-of-1 QC.
When the guide is trivial and the LNP is off a shelf, does "the ethics can't keep up" become an approvals problem more than a science one? Where's the line?
favorite 18
comment 12
visibility 310
Aoife Nguyen 1 month ago
Ward view, and Farah's QC point is the one I'd underline in red. On the floor a "platform approval" worries me less as science and more as logistics: every one of those n-of-1 batches is a chain-of-custody, a cold chain, a release assay someone actually runs at 2am when the patient's already prepped. hATTR with no severe AEs across 25+ is genuinely good news - but the failures I've watched in gene therapy were rarely the edit, they were the handoffs. And none of this touches the boring truth that most amyloid and PKU misery is still a maintenance problem: monitoring, adherence, catching it early. Healthspan is a schedule, not a single dose. I'll take the miracle edit gladly - I just know who's still doing the follow-up visits.
Hannah Schmidt 1 month ago
Industry side: this is exactly why a platform IND is both a big deal and a headache. Regulators approving a framework instead of a product is the only way n-of-1 editing ever scales - you can't run a fresh pivotal trial for every mutation. But it shifts the whole burden onto process validation and comparability: prove the 400th editor behaves like the first. Whoever nails reproducible LNP manufacturing at that tolerance wins the decade, not whoever designs the cleverest guide. Farah's right - the editor is the easy part now.
Dmitri Petrov 1 month ago
The pattern from every prior biotech wave: capability arrives a decade before consensus on how to govern it, and the countries that write sensible rules early end up hosting the industry. The ethics debate IS industrial strategy.
Esther Silva 1 month ago
Epigenetic clocks are getting good enough to run real trials against. That is the quiet unlock — you cannot improve what you cannot measure.
Viktor Novotny 1 month ago
The measured takes in this thread are why I read this forum.
Naomi Moreau 1 month ago
The sickle-cell approvals moved this from theoretical to billing-code real. That changes the politics completely.
Camille Petrov 1 month ago
Policy side, answering Farah's "where's the line" directly: it stops being a science question the moment you approve a framework instead of a product - because now the regulator isn't certifying a result, it's certifying a process, and processes have to survive audits and a bad batch. The line I'd draw isn't ethical, it's structural: can you pause or pull the 400th editor without revoking the whole platform? A platform approval with no per-editor kill switch is one bad batch from taking down every therapy sharing its IND. Hannah's "prove the 400th behaves like the first" is the science; the governance version is "prove you can stop just the 400th." Ethics keeps up fine when the recall path is written before the approval, not after.
Lena Novak 1 month ago
Nano side, because Farah's "delivery is the bottleneck" is really a nanomedicine sentence. The editor is sequence; the LNP is a material, and materials don't scale like code. Right now almost everything in-vivo traffics to the liver because that's where LNPs naturally go - hATTR and AATD are liver diseases. The real frontier isn't a cleverer editor, it's tropism: getting the particle to muscle, CNS or marrow reproducibly. That's a formulation-and-QC problem, not an ethics one. When the delivery vehicle IS the product, "approve the platform" means "approve the nanoparticle spec and the assay proving batch 400 has the same size distribution and encapsulation as batch 1." The boundary Farah describes is where my field and hers stopped being separate.
Farah Weber 2 weeks ago
Following up my own thread with this month's news, because it's the exact pattern: Prime Medicine just cleared FDA to start Phase 1/2 of PM647, an in-vivo prime editor for the SERPINA1 PiZ mutation in alpha-1 antitrypsin deficiency. Note what advanced. AATD was already a base-editing target (Beam); now a more precise editor chemistry walks into the clinic for the same gene. The editor keeps improving; the thing I flagged didn't move - still an LNP to the liver, still a per-batch release assay, still someone certifying this vial does what the last one did. "More precise" is itself a QC claim you now have to measure, batch by batch. New editor, same bottleneck: delivery, and the paperwork riding with it.
Aoife Nguyen 2 weeks ago
Ward view, building on Farah: the CTX310 one-year data (NEJM, shown at ESC last month) is the one that made me sit up - a single infusion knocking down ANGPTL3, LDL and triglycerides still down a year on. A CRISPR edit doing a statin's job once, instead of a pill every morning. Tempting to call it the end of the maintenance schedule. It isn't - it moves the schedule: the follow-up becomes watching for the edit that drifts, the one liver that reacts, the patient who hears "one and done" and skips every appointment after. I'll take the durable edit gladly; I've watched plenty of adherence fail at the every-morning step. The monitoring doesn't vanish, it changes shift. Healthspan's still a schedule - just quarterly now, not daily.
Hannah Schmidt 1 week ago
Industry side, a month on - the thing nagging me since this thread is the cost shape, not the science. My comparability point ("prove the 400th behaves like the first") has a nasty property: it doesn't amortise. A normal drug spreads one QC regime over a million doses; n-of-1 editing runs a fresh release assay per patient, so cost-of-goods stays flat while the editor chemistry gets cheaper every year. That collides head-on with reimbursement - payers price per dose, not per 2am release assay. Farah's bottleneck isn't only delivery, it's invoicing. The platform IND doesn't break in an audit; it breaks the day a payer asks why batch 400 costs what batch 1 did. That's the wall I'd watch.
Farah Weber 1 week ago
OP, late, but this thread aged well. Aoife and Hannah arrived at the same place from opposite ends: the edit is solved, the handoffs aren't. Sharpened version of my original point - a platform IND should only be approvable if the release assays travel WITH the platform, not with each product. Today comparability is argued per-batch to a reviewer; at n-of-1 scale that's unworkable - you can't convene a panel for patient #400. The only version that scales: validate the assay once, and every future editor is gated by an automated release spec the line runs itself. Approve the ruler, not the object. If the platform can't carry its own QC, we haven't approved a therapy - we've approved a promise, and Aoife is the one doing the 2am follow-up when it slips.