Episode 162 - November 7, 2025

Summary

• The week’s defining event was George Tidmarsh’s departure as CBER director, raising questions about Vinay Prasad’s influence at the FDA. Adam Feuerstein’s reporting: Tidmarsh walked into Marty Makary’s office expecting to hear Prasad was fired — instead he was the one getting the boot, centered on healthcare investor Kevin Tang’s complaint that Tidmarsh used the post to “exact revenge.” Brian Skorney’s worry: it may now be “more of a Vinay Prasad FDA than a Tidmarsh FDA,” with review teams influenced by leadership tone rather than explicit orders.
• UniQure’s Huntington’s setback is the shock that reprices the whole single-arm gene-therapy complex. Paul Matteis details how the company aligned with FDA on a statistical analysis plan four months ago, got breakthrough designation on its 2-year data, delivered better-than-expected 3-year results — 80% disease slowing versus natural history at 2 years and 75% at 3 years for the high dose — and was then told the natural-history analysis was exploratory and could not support a BLA. “Are we supposed to believe that was a sham meeting?”
• Sarepta’s ESSENCE failure produced a wake-up moment for longtime observers. Graig: the largest, longest DMD study to his knowledge showed no effect — “I don’t think it worked.” Adam went “full snark”: the 6-minute walk test failed, then North Star, then the stair-climb endpoint — “These drugs just don’t work… all we’re hearing are excuses for why we should do something else.”
• The panel’s core complaint isn’t rigor, it’s unpredictability. Paul is “open to” the theoretical case for tougher efficacy standards, but investors now doubt whether any reported FDA alignment is real — on Dyne’s year-end placebo-controlled DMD readout, with 15 times as much dystrophin in Phase 2, “what is the bar on the functional data? You don’t know and I don’t know.” Late-September’s bullish cell/gene draft guidance versus the actions since — “they just don’t line up.”
• Biohaven’s troriluzole CRL is the transparent counterexample: the CRL hammers the natural-history statistics, with tipping-point analyses that fail. Brian notes the Phase 3 study missed with a p-value above 0.9 before the open-label-versus-natural-history pivot — unlike UniQure, “we’re less blind” here. Stoke sold off in sympathy: Yaron calls zorevunersen’s 60%–80% seizure reduction in Dravet “unquestionable,” but the early-filing bet that essentially doubled the stock now hinges on an opaque December FDA meeting, with Phase 3 data otherwise not due until the second half of 2027.
• On the flip side, the Metsera bidding war is good for biotech. Chris Garabedian: after five years of a buyer’s market, “pharma is tripping over themselves to get good assets” — the Pfizer/Novo sums are “rounding errors” for the acquirers. Another panelist said patent cliffs should sustain M&A, while a venture-side panelist called venture “probably the most optimistic it’s been,” though “not all-in.”
• Arena BioWorks’ sudden shutdown underscores skepticism about its privatized-discovery model, and the commercial tape stays unforgiving. Chris questioned Arena’s $500M model and its planned $50M-a-year spending when seed/Series A funding is at a 10-year low. Soleno beat a street-high estimate ($66M versus $58M) and still fell more than 30% on Skyclarys/Daybue-style plateau fears — compounded by management citing the Scorpion short report on the call, contrary to “management 101: do not publicly call out the shorts.”
• Watch Terns into ASH: its allosteric CML TKI posted 64% six-month MMR in the abstract versus a 25% benchmark for another allosteric drug and 32% for Enliven’s compound, and the stock has more than doubled since the beginning of the week — small dataset, update in early December in Orlando. In orexin, Centessa’s data validate the class in narcolepsy type 2 but “didn’t meet the bull case”; Alkermes’ bigger, longer NT2 Phase 2 is the catalyst that could size the whole space.

Deep dive

1. Tidmarsh’s exit raises questions about Prasad’s influence at CBER

• Adam’s backstory: a week before the departure, Tidmarsh was called into Makary’s office thinking he’d hear Vinay Prasad was being fired from the FDA — “instead he learned that it was him who was getting the boot,” centered on healthcare investor Kevin Tang’s complaint that Tidmarsh was using the CBER post to “exact revenge” over an old falling-out. Put on leave, resigned, fought the resignation — “but now he’s gone.”
• Yaron’s mea culpa: “I have mud on my face — I thought George Tidmarsh was one of the better picks because of his biotech experience. And man, what a rookie mistake,” one that hurt agency credibility by raising questions about whether personal preferences affect how the FDA treats companies and CEOs. As of today there is no CBER director; people are reporting that Peter Marks has turned down the role, as has the OMD director.
• Brian’s investor read: the concern layered on top is that the uniQure decision may be “tied somewhat to Vinay being more of the emerging victor in the civil war” — and Prasad is someone who, based on what Brian knows of him, would “look at and scoff at” the flexible-approval messaging.

2. UniQure: an aligned SAP, a breakthrough designation — then the door closed

• Paul’s reasoning chain, worth keeping in full: UniQure’s regulatory alignment looked “more thorough and rigorous than average” — a meeting a year ago on the natural-history control, a spring meeting that aligned with FDA on an SAP, including the database and methodology, and breakthrough designation alongside the 2-year data. Then 3-year data came in better than expected — 80% disease slowing versus natural history at 2 years and 75% at 3 years for the high dose — and FDA conveyed that the analysis was exploratory and could not support a BLA. “Are we supposed to believe that was a sham meeting?”
• Adam’s pushback: aren’t these review issues — sensitivity analyses across Enroll-HD, TRACK-HD and another similar cohort are the kind of issues to deliberate during review or at an AdCom, not grounds to “just tell a company no, you can’t file.” His prescription: “radical transparency… can work both ways” — if UniQure thinks this is a unilateral power grab, put all the meeting minutes and agreements out in the open.
• Brian’s mechanism: probably not Prasad explicitly stepping in — “review teams feel emboldened one way or another based on the leadership mentality.” Marks pushed flexibility and showed Wilson Bryan the door in an effort to be more flexible; Prasad’s views on natural history in slowly progressive diseases are well publicized on his YouTube channel and podcast. UniQure says Prasad was not in the meeting. Paul wonders whether the clinical team had been ambivalent but could not convey that under the previous CBER leadership, or whether the teams are now being directed separately.
• Paul’s half-joke on the next catalyst for the stock: “let’s see the STAT News exposé and what’s happening at the FDA.” Adam: “We’re trying, Paul.”

3. The dissonance: flexible guidance, inflexible actions

• The panel questioned the gap between the leadership listening tour and late-September cell-and-gene-therapy draft guidance — which read as flexible and aligned with what UniQure and other companies are doing — and the recent actions. “Are we supposed to believe the public commentary is a little bit more like politics?” Paul agreed that “it’s the dissonance… they just don’t line up.”
• Adam’s history lesson: this isn’t new — for decades the hierarchy signaled flexibility, including Janet Woodcock’s view that a small patient study could support approval with a sufficiently large effect size, while divisions applied old-school rigor. The hierarchy’s job “is almost promotion”: presenting the FDA as willing to work with industry and not getting in the way of patients.
• Graig’s “Prasad parachute” frame: “You can go on PubMed and type in ‘Prasad parachute.’” Prasad accepts a 10-patient approval for a true medical parachute with a 100% effect size — “but his whole career arc is showing how most medical interventions are not actually parachutes.” Nobody looks at the UniQure data and sees “a Lazarus effect”; these are bubble applications, and leadership tone may tip them.
• Adam’s concrete casualty: Ultragenyx said this week that the UX111 filing plan in MPS IIIA, agreed under Peter Marks’ CBER, shifted under the new FDA — “the rug was pulled out from under them” with a CRL out of nowhere, because the FDA wanted manufacturing documentation settled ahead of time rather than during review.

4. Sarepta’s ESSENCE failure: the old thesis gets tested

• Graig: alongside earnings, Sarepta disclosed failure of the largest, longest DMD study to his knowledge — for the two follow-on exon-skippers, golodirsen and casimersen — with the company arguing that excluding COVID patients showed more favorable trends. His change of mind: “It’s kind of a wake-up moment for me… I don’t think it worked.” He had pushed for two years precisely because a slowly progressive disease might hide in a 50-patient, one-year study; at this size, no effect “calls into question what are we doing here.”
• Adam went “full snark” on X during the call: the 6-minute walk test was the best endpoint until it failed, then North Star, then stair climb. “These drugs just don’t work… all we’re hearing are excuses for why we should do something else.”
• Chris, who got the original ESSENCE master protocol approved as Sarepta CEO, opposed the two-year design and is “appalled” that a study designed in 2014 read out only now. He thought the drug would show an effect but now thinks it hasn’t. His asymmetry argument: chronically administered oligos carry clearer safety than the gene-therapy datasets, and low-bar approvals did not quell innovation — “we’ve only seen more and more technologies.” But a mediocre gene therapy may prevent patients from receiving a better one later because of immunogenicity. The endgame is murky: after more than 10 years and thousands of patients, does FDA really pull the drugs, or do insurers refuse reimbursement while “these poor families” pay out of pocket?
• The historical frame: eteplirsen was the first intervention by a CBER director to override Ellis Unger’s recommendation for a CRL. Yaron said Califf supported the Scientific Dispute Resolution Board, which favored Janet Woodcock’s decision, and that the dispute-resolution process was properly followed. Yaron contrasted that with AVXS-101: 12 type-1 SMA patients still living after two years, a “gold standard” parachute. He had always said DMD was not that.

5. Rigor is only bullish if it’s predictable

• Paul’s bottom line on whether the new regime is good for the industry: “It can’t be good right now unless it’s predictable… the market’s telling you that was not predictable.” He’s open to the theoretical argument that more efficacy rigor could be better — what he hates is that investors now ask whether reported FDA alignments are real at all.
• His live example: Dyne’s placebo-controlled DMD data by year-end, with 15 times as much dystrophin in Phase 2 — “what is the bar on the functional data for Dyne to get approved? You don’t know and I don’t know.” Investors who want to fund programs built around regulatory flexibility are having trouble deciding whether those agreements can be trusted.

6. Biohaven’s CRL: the case where we can actually see the review

• Brian’s walkthrough: troriluzole, reviewed by CDER rather than CBER, failed its Phase 3 study with a p-value “above 0.9,” then pivoted — after a long documented history of FDA meetings on an SAP — to comparing three-year open-label-extension data against a matched natural-history cohort. The analysis hit its primary and all secondary endpoints and was marketed as potentially “the gold standard” for a natural-history dataset.
• The CRL shows why it lost: unaccounted selection biases, and tipping-point analyses around the assumptions that “wind up failing statistically.” Contrast with UniQure: “we’re a little blind on the UniQure discussion; we’re less blind on Biohaven” — this was a review in which the effect size could not be trusted. Biohaven criticized FDA in its own release, and the stock got hammered.

7. Stoke: the early-filing dream, repriced by UniQure

• Yaron’s setup: zorevunersen upregulates Nav1.1 in Dravet, with four years of data showing 60%–80% seizure reduction — “unquestionable” — in patients who had been on six previous drugs, plus neurocognitive benefit; consultant checks call it “probably the best-looking product in development ever for Dravet.” Dog toxicity capped U.S. dosing, a partial clinical hold followed an NHP finding never seen in patients, and the U.S. study, U.K. study and Phase 3 program use different doses — “which dose do you even suggest to put on the label?”
• The trade: Phase 3 data are not due until the second half of 2027, so new CEO Ed Kaye’s discussion of filing on Phase 1/2 data under breakthrough designation essentially doubled the stock. At roughly $38 a share, investors were thinking about a $1.5B market cap, $400M in cash, a roughly $500,000-per-year drug and potentially $2B in peak sales — the chance to reach market in the second half of next year rather than the second half of 2028.
• December’s FDA meeting has roughly three scenarios: a clear go-ahead, a weak “you can file if you really want to” signal, or the gray middle in which filing is a review decision. That last message is “completely opaque” to Wall Street.
• Adam wonders whether investors “just got too far ahead of themselves”; the irony another panelist flags is that outgoing CEO Ed Kaye was the CEO at Sarepta when Exondys 51 was approved. “Small world.”

8. Metsera’s bidding war and Arena’s collapse: a barbell market

• Chris on Metsera: after five years of a buyer’s market in which investors dictated startup terms, the Pfizer/Novo fight is “a clear signal… pharma is tripping over themselves to get good assets” — the sums are “rounding errors” against the acquirers, and even the mudslinging, including Novo calling Pfizer’s antitrust claim “BS,” is “fun to watch.”
• Another panelist said patent cliffs are real and should support continued M&A for assets with clear clinical proof of concept, citing Avidity as a recent example, alongside the XBI above 100, follow-ons and an IPO window.
• Adam on Arena BioWorks, reported by colleagues Jason Mast and Allison DeAngelis: the buzzy $500M academia-industry bridge shut down suddenly at an all-hands meeting, essentially because of the difficult financing market for private companies.
• Chris said many people viewed the model as dubious: privatizing leading science from the Broad, Whitehead and Koch institutes and spending “$50 million a year over 10 years” to spin out biotechs, when seed and Series A investment is at a 10-year low and money flows to de-risked B, C and D rounds.
• A venture-side panelist said venture is a lagging indicator and “right now is probably the most optimistic that venture has been,” but “we’re not all-in.” Another quarter or two of IPOs and steady valuations could bode well for private companies receiving money in 2026, with best-in-class programs funded before unvalidated novel targets.

9. Soleno’s beat-and-drop, orexin sorting, and Terns into ASH

• Brian on Soleno: the Prader-Willi drug’s second quarter doubled the first — $66M versus his street-high $58M — and the stock still fell more than 30%. Scorpion Capital’s short-report thesis, involving hypoglycemia, edema, hospitalizations and discontinuations, “doesn’t get a lot of traction” with investors, and the data do not show patients and physicians abandoning the drug: there were 397 new-start forms, roughly 30 per week, and discontinuation rates looked normal. But investors fear a Skyclarys- or Daybue-style plateau in the $80M–$100M range. He likes the non-promotional management, but citing the short report on the call broke “management 101: do not publicly call out the shorts.”
• Paul on Centessa’s orexin readout: the drug clearly works in narcolepsy type 1 with a big effect; the NT2 effect is clinically significant but “maybe didn’t meet the bull case,” looking similar to peers — and smaller Phase 1-type sleep datasets can “regress a little to the mean” in Phase 2 and Phase 3. Still, the data validate the class: pushing exposures higher can produce success in NT2, and Alkermes’ bigger, longer NT2 Phase 2 is the catalyst that could determine how investors view Centessa and size the class — “you can start to dream the dream around other sleep disorders.”
• In the Terns discussion, a panelist described the company’s allosteric-site-targeting CML TKI’s 64% six-month MMR in its ASH abstract versus a 25% benchmark for another allosteric drug and 32% for Enliven’s compound — “really a home run.” The stock has more than doubled since the beginning of the week, with the caveat of a small dataset and an update at ASH in Orlando in early December.