Hugo Vance: You seemed distracted when you walked in — something on your mind before we start?
Lila Soto: Yeah, I mean — I've been trying to figure out how to describe this to someone who's never heard of it. Like, imagine a pharmacist, but instead of mixing pills from a shelf, they're rewriting a single broken sentence in your body's instruction manual. And they build that sentence specifically for you, no one else.
Hugo Vance: That's the analogy you're reaching for with KJ Muldoon.
Lila Soto: Exactly him. May 2025 — a critically ill infant, a CPS1 gene variant that basically no one else on the planet shares, and a team that went from mapping his specific mutation to a completed personalized mRNA-CRISPR therapy in a matter of months. They called it kayjayguran abengcemeran. K-abe.
Hugo Vance: Months. I want to sit with that number. That is not how drug development works.
Lila Soto: It's not how it worked, no. That's kind of the whole story. Because what k-abe is — the core idea, one clean sentence — is: we can now build a gene therapy for one specific person's one specific mutation faster than the regulatory system knew how to handle. And then the question became, what do you do with that?
Hugo Vance: And the FDA's answer to that question is what's driven everything since.
Lila Soto: One baby. One emergency. And then something much bigger started to follow.
Hugo Vance: And what followed — the FDA's answer — that's where I'd slow down, because it's more complicated than it looks. Peter Marks, when he was still running the biologics division, was sketching this out publicly in 2024. The Plausible Mechanism Pathway, the Platform Technology Designation program — he wasn't waiting. He was rehearsing the architecture out loud.
Lila Soto: Rehearsing it — meaning the guidance didn't come first?
Hugo Vance: The guidance came last, actually. August 2026, the FDA issued a draft letting sponsors reuse validated CMC and clinical data across related genome-editing programs. October 27th, they published the Individualized Human Genome Editing Products draft — the one that says a single BLA could cover multiple patient-specific variants inside a defined 'individualized feature space,' no new clinical trial for each. And then November 2nd, the Potency Assurance guidance dropped — the one piece of that whole cluster that is actually final, not draft. But September — one month earlier — Andelyn Biosciences had already announced the Curator platform produced Ultragenyx's Fayuvi for Sanfilippo syndrome type A. First commercial gene therapy milestone on that platform.
Lila Soto: Hold on. Fayuvi was approved before the guidance dropped?
Hugo Vance: September 18th, 2026. The guidance cluster: late October, early November. So manufacturers were already operating in this modular space, and then the FDA wrote the rules around what was already happening.
Lila Soto: So the story isn't 'FDA leads the way' — it's more like, the institution caught up to its own experiments.
Hugo Vance: Yes. And I'd be cautious about celebrating too quickly, because — well, the 'individualized feature space' concept, which is the core of the whole October 27th guidance, hasn't been adjudicated under real commercial pressure yet. Someone will eventually ask: is a novel heterozygous mutation in the same gene a variant inside the space, or a new indication entirely? And the honest answer right now is — the FDA doesn't have a worked example. It's framework language.
Lila Soto: And most of this is still draft. Like, the Potency Assurance piece is final, but the big structural one — the BLA covering multiple variants — that's still out for comment.
Hugo Vance: Indeed. One final guidance, two drafts, and the concept that holds the whole thing together is untested. That's the actual state of it.
Lila Soto: But okay — untested framework is one thing. I keep getting stuck on something more concrete than that. Like, can anyone actually *make* these therapies at scale? Because Andelyn Biosciences and the Curator platform — September 18th, first FDA-approved commercial gene therapy manufacturing milestone, Fayuvi produced in Columbus, Ohio — that's real. That's not theoretical.
Hugo Vance: It is real. And that's exactly what makes the next part harder to say.
Lila Soto: One CDMO. One therapy. One facility. I mean — picture a family in rural Montana whose child has a variant of Sanfilippo syndrome type A that's close to what Fayuvi covers, but not quite. Not the exact mutation. The cold chain alone from Columbus to a hospital in Billings — actually, I don't even know if Billings has the infrastructure to receive it. That's before we touch per-patient cost.
Hugo Vance: Modular regulatory approval is not the same thing as modular manufacturing. The FDA tidied the paperwork. The cold chain is still a mountain.
Lila Soto: And ARPA-H knows it — that's the tell. September 2026, they announced $125 million for the GIVE program. Genetic Medicines and Individualized Manufacturing for Everyone. The whole point is automated, distributed, multi-site biomanufacturing for individualized RNA-based genetic medicines. The government is funding that because the infrastructure cannot yet support the volume or geography the regulatory framework envisions.
Hugo Vance: And we should be honest — GIVE is a program, not a proven solution. The $125 million is there. The distributed factories are not.
Hugo Vance: You see, this is the pattern I keep returning to. Orphan drug law, 1983 — promised accessibility, created a market, left manufacturing and reimbursement entirely unsolved. We're watching the same displacement happen. The regulatory permission arrives. The structural problem migrates.
Lila Soto: And the reimbursement side — that's genuinely uncomfortable, because CMS expanding the Cell and Gene Therapy Access Model sounds like a signal, but it's still a plan. And historically, that gap between approved and actually paid for is where families disappear. We'll get to why that matters in a minute — but it's the part that might undo everything we've just described.
Hugo Vance: And 'plan' is doing a lot of work in that sentence. October 27th, 2026 — the same day the FDA published the Individualized Human Genome Editing Products draft guidance, CMS announced plans to expand the Cell and Gene Therapy Access Model beyond sickle cell disease. Same day. That looks like coordination. But read the word carefully. Plans.
Lila Soto: Two drafts talking to each other.
Hugo Vance: Essentially, yes. And I don't want to dismiss the CMS signal — it's real, it matters. But a plan announced alongside a draft guidance is not a payment model. It's an intention sitting next to a framework. Neither has cleared.
Lila Soto: And historically, that gap — I mean, approved therapies have literally sat on shelves because no payer would cover them. That's not hypothetical, that's the pattern.
Hugo Vance: Orphan drug law, 1983. Same structure. The regulatory door opened. The economics never followed. Forty years.
Lila Soto: But if the FDA approves therapies for, say, fifty genetic variants of one rare disease, and a payer covers five of them... the other forty-five patients got nothing from all of this. The regulatory acceleration meant nothing to them.
Hugo Vance: And the decision about which five — that lands on a hospital ethics committee. A specific room. Someone actually has to look at ten eligible patients and say: we have payer coverage for three this quarter. Who waits.
Lila Soto: And that decision — the regulatory framework doesn't touch it at all. That's the gap. Legally possible. Actually feasible. Those are still two completely different places.
Hugo Vance: And that gap — legally possible, actually feasible — that's where KJ Muldoon still lives. Kayjayguran abengcemeran worked. One child, one mutation, months of work. But it hasn't become the template yet. It's still the exception.
Lila Soto: I keep thinking about your pharmacist analogy — well, my analogy, the one I walked in with. Rewriting one broken sentence in one person's instruction manual. That image was always about possibility. And it still is. But now I'm sitting with the part underneath it — the factory that makes the ink doesn't exist at the scale we need. The individualized feature space is untested under commercial pressure. CMS has a plan, not a commitment. The FDA built a permission slip. A real one. That's not nothing.
Hugo Vance: No. It's not nothing. Peter Marks was sketching this in 2024. The institution moved. That is genuinely rare. But the factory that makes the therapy — it doesn't exist yet. And the payer hasn't said yes.
Lila Soto: So KJ Muldoon is still the exception.