Episode 148 - July 18, 2025

Summary

• The week’s defining story was Sarepta’s credibility collapse: 12 hours after a well-received restructuring call (36% of the workforce cut, gene therapy de-emphasized, a $500M revenue “floor” pitched), reporters — not the company — revealed a patient death in the limb-girdle program using the same rAAVrh74 vector, reportedly known to the company for perhaps a month and not answered when an analyst asked point-blank about safety. Adam Feuerstein called it “a lie by omission,” noting CEO Doug Ingram insisted it was “not a material event” and an “NPV-based decision” — and the episode closed with the FDA requesting that Sarepta voluntarily halt all Elevidys shipments, with shares halted.
• Feuerstein’s broader diagnosis frames the whole episode: biotech “has a serious credibility and accountability problem,” and until analysts call problems out, investors stop accepting them, and boards and executives are held accountable, generalists will treat the sector as “some niche curiosity that people just laugh at.” From outside, the Sarepta saga looks like “a raging dumpster fire attached to a clown car”; Fazeli added that XBI’s 3-, 5- and 10-year returns make capital allocators ask why they should invest in a sector that has not really moved in 10 years.
• The Ultragenyx CRL for its Sanfilippo gene therapy was purely CMC — including a CRL-listed deficiency over having five temperature probes in a shipping container when FDA wanted seven — pushing approval from August to possibly 2026. Yaron Werber flagged the bigger tell: CEO Emil Kakkis detects a shift at FDA away from the heparan sulfate biomarker (the Peter Marks accelerated-approval program) toward neurocognitive endpoints, and the CRL landed 5-6 weeks before PDUFA without the full package apparently reviewed — “almost seems that there was a political reason.”
• GSK’s Blenrep took “a bruising ODAC” — DREAMM-7 and DREAMM-8 voted down 7-1 and 5-3 despite statistically significant superiority with survival trends — over dosing GSK was told to lower and only 5% U.S. enrollment versus CARTITUDE-4’s 15%. Sam Fazeli estimates the drug is ~30% of GSK’s growth to 2030; Werber tied it to a new FDA/ODAC posture against outdated ex-U.S. control arms (Columvi’s second-line DLBCL CRL, with a CRL landing “literally now”): “I’m glad they’re not getting approved.”
• Market read: Q2 venture shifted decisively away from early-stage — Series B-and-later rounds more than 2x seed/Series A, zero IPOs — yet a forum of ~80 CEOs that heard debriefs from FDA leadership meetings “universally all walked away with a real feeling of encouragement.” Sam’s H1 M&A tally: 12 large-biopharma deals worth $30.7B (Blueprint plus Intracellular Therapies ~$23B; ~$40B with Verona), and Werber countered Josh Schimmer’s worry that Merck’s $10B Verona buy shrinks the bid — debt capacity runs to several hundred billion and $10B is under a year of Keytruda cash flow, with Keytruda’s IRA date pushed from 2028 to 2029. Schimmer is constructive into year-end: “most of the bad news is already priced in,” implying a ~10% move up.
• Kailera/Hengrui’s HRS-9531 posted 17.7% weight loss in Chinese phase 3 at 48 weeks versus Zepbound’s 17.5% at ~52 weeks in comparable Chinese patients — but with the curve “still continuing” where Zepbound plateaued at 44 weeks, an 8-mg dose still being tested, and target-dose exposure nowhere near FDA’s 52-weeks-at-target-dose requirement. Werber says the tolerability “definitively looks better” than Zepbound across studies; a 4,500-patient phase 3 could start by year-end, though a U.S. partner and more capital look necessary.
• In Sam’s 53-physician U.S. lung cancer survey, AstraZeneca’s AVANZAR led as most anticipated practice-changing trial (49%), Summit’s ivonescimab drew ~40%, and BMS’s relatlimab combo was the surprise entrant, while BioNTech’s BNT327 scored low on familiarity (only ~5-7 U.S. centers). On earnings: J&J cut its tariff-impact estimate from $400M to $200M with Spravato up 30% quarter-over-quarter; Novartis beat and raised but trimmed the 2025 Cosentyx forecast (12% of 2024 sales, an $8B peak-sales estimate before 2029 patent expiry) and launched a $10B buyback.

Deep dive

1. Biotech’s problem isn’t biology — it’s the self-inflicted wounds

• Schimmer’s opening provocation: is biotech “too challenging to ever be anything other than domain for real experts”? Werber’s answer depends on the era and cap size — large caps trade like pharma — but small/mid-cap today “is not for the faint of heart” given volatility and FDA uncertainty.
• Feuerstein’s thesis, delivered before the Sarepta discussion even started: “the sector has a serious credibility and accountability problem,” and until analysts and investors self-police and boards hold executives accountable, biotech stays “some niche curiosity that people just laugh at.” Everyone accepts that biology is hard — “it’s the self-inflicted stuff… that keeps happening all the time” that must be cleaned up.
• Garabedian’s comparison: tech can offer good returns and is more tangible and understandable; Fazeli’s addendum is the capital-allocation problem — look at XBI’s 3-, 5- or 10-year returns, and with rates still high, investors ask why they should commit capital to a sector that has not really moved in 10 years.

2. Sarepta: the death nobody disclosed — and the shoe that dropped on air

• Schimmer’s setup: the restructuring (36% of staff, ~500 people, retreat from gene therapy, a $500M gene therapy revenue floor) initially sent the stock up on debt-overhang relief — then BioCentury, followed by Endpoints, reported a death in the limb-girdle program using the same rAAVrh74 vector, in an adult patient, apparently known to the company for upwards of a month and unaddressed even when an analyst asked directly whether safety drove the limb-girdle shutdown.
• Feuerstein’s crucial correction: Sarepta didn’t announce anything — “unless there were reporters out there pursuing these kinds of stories, there wouldn’t have been a disclosure.” Ingram called it “not a material event,” an “NPV-based decision” — “it’s a lie by omission… how do you evaluate a management team like that going forward?” And on the day 500 employees were laid off, the company announced executive promotions and raises: “that does not sit well.”
• Garabedian, as former Sarepta CEO, on why controversy has surrounded the company: “the FDA is not a monolith” — he cycled through three division directors during eteplirsen, a drug with “arguably a small data set that was not unequivocal” — and his own approach was radical transparency, publishing FDA letter quotes in press releases. Feuerstein drew a through-line to the PPMD conference about a year ago, where a DMD mother’s public criticism was harassed and her comments edited out of the video at Sarepta’s request: “you can draw a through line to that.”
• The episode literally ends where it opened: headlines cross that FDA is requesting Sarepta voluntarily stop all Elevidys shipments, shares halted for volatility.

3. The counterexample, and the governance fix

• Schimmer’s deliberate juxtaposition: Amylyx’s well-received ENDO webinar for its GLP-1 inhibitor in post-bariatric hypoglycemia matters less than who’s running it — co-CEOs Josh and Justin, who pulled their ALS drug from the market when the trial failed, as promised. Garabedian: “we need more management teams like that.”
• Werber’s structural point: the FDA publicly posting CRLs is something “we need badly,” because there is no comprehensive disclosure rulebook — no “Gandalf… with a book of 3,600 pages” appears when you become a Section 16 officer. Disclosure is left largely to risk-management, board, general counsel, CEO and CFO discretion: “that’s absolutely not okay. This is really a governance issue.”

4. Ultragenyx’s CRL: seven thermometers, and a quiet shift on accelerated approval

• Via Feuerstein’s Readout Loud interview: the Sanfilippo gene therapy CRL was entirely CMC — no clinical shortcomings — including a listed deficiency because the shipping container had five temperature probes and FDA wanted seven. “Nothing to do with the drug, nothing to do with safety or efficacy.” Kakkis was careful not to blame Makary or Prasad — this is red tape “built in to the FDA right now” — but approval slips from August to possibly 2026, “a significant meaningful amount of time” for these patients.
• Werber’s zoom-out: this is one of three drugs up for biomarker-based accelerated approval at CBER — “the Peter Marks program” — and Kakkis is “definitely detecting” a move away from heparan sulfate as the approvable biomarker toward neurocognitive endpoints. Plus the CRL came 5-6 weeks before PDUFA without the full data apparently reviewed: “it almost seems that there was a political reason.”
• Schimmer’s skepticism — worth keeping: a single-arm, small, open-label, biomarker-endpoint study amid Vinay Prasad’s questioning of accelerated approval and disagreements with Marks means the CMC fix may not be the whole story. “I still want to see that this gets the ultimate approval based on this smaller data set.” Feuerstein, on whether fear of retribution is up: “hard to say,” but cites internal ousters, an internally mentioned departure whose name was unclear in the transcript, and the KalVista situation.

5. Blenrep’s bruising ODAC signals the end of the ex-US shortcut

• Fazeli’s readout: DREAMM-7 and DREAMM-8 voted down 7-1 and 5-3 despite statistically significant superiority with survival trends — FDA said the dosing and tolerability were wrong, had told GSK along the way to drop the 2.5-mg start, and Peter Marks questioned the 5% U.S. enrollment versus CARTITUDE-4’s 15%. At ~30% of GSK’s estimated growth to 2030, “that’s why the stock’s taking a bit of a hammering.” Feuerstein said physicians he spoke to believed, “we know how to manage this drug. We want it on the market” — and wished GSK “had done a better job with it.”
• Werber’s read — a new FDA/ODAC posture: like Columvi’s second-line DLBCL CRL (outdated STARGLO control arm, few U.S. patients — and a CRL landing “literally now”), companies are running superiority trials against comparators nobody uses anymore. “Companies are not really trying to innovate. They’re trying to manage risk… I’m glad they’re not getting approved.”

6. Venture winter persists — but CEOs left the Makary meetings encouraged

• Garabedian’s report from a private forum of ~80 CEOs who heard debriefs on FDA leadership meetings across New York, Boston at BIO, San Francisco and D.C.: attendees not sympathetic to the administration’s politics “universally all walked away with a real feeling of encouragement” — active note-taking, consistent messaging across venues, and an apparent commitment to make early-stage U.S. trials easier (Chris and other VCs have been moving early studies offshore because FDA has been “too conservative about getting drugs into clinical testing”).
• The Q2 data cuts the other way: after a Q1 in which biopharma seed/Series A dollars actually exceeded Series B-and-later, Q2 reversed hard — later rounds more than 2x early-stage — matching what Chris sees on the ground: seed-to-A is “really really hard.” Zero IPOs in Q2; M&A ticked up. HSBC’s report (the former SVB report led by John Norris) shows oncology dropping while neuro/CNS, ophthalmology, I&I, cardiometabolic and respiratory broaden the mix, with every VC auditing runway to “that elusive clinical proof of concept data set.”

7. M&A firepower is real; sentiment cautiously turns

• Fazeli’s H1 tally: 12 large-biopharma takeouts worth $30.7B — Blueprint and Intracellular Therapies alone ~$23B, ~$40B adding Verona — versus 2022’s $73.6B and 2023’s “monster year of 121 billion”; two or three big second-half deals could put the year within reach of the second-best year and at least meet or beat the 2020 figure. Oncology retook I&I in deal count at roughly a third.
• Schimmer’s contrarian question: Merck just allocated $10B to Verona — shouldn’t every other stock trade down a nudge as a buyer leaves the table? Werber’s rebuttal: large biopharma debt capacity runs to several hundred billion, $10B is under a year of Keytruda cash flow, and Keytruda just got an extra IRA year (2028→2029). Garabedian adds that deals ignite whole disease categories: “the positives outweigh the negatives.”
• Year-end calls: Schimmer — the sector is +6-7% over the last month but still -10% YTD; his history-based framework implies a ~10% move up by year-end, and “most of the bad news is already priced in.” Schimmer linked Summit’s unexplained 16% pop to AstraZeneca deal expectations (“let’s see. Maybe Monday”). Feuerstein’s quip: “If I had a nickel for every time I was optimistic for the outlook of biotech and was proven wrong, I would at least have a dollar.”

8. Data and earnings: a Zepbound-matcher from China, lung cancer surprises, mixed pharma prints

• Kailera/Hengrui’s HRS-9531 (GLP-1/GIP, a Bain-Atlas spinout run by Ron Renaud, with Scott Wasserman of Repatha pedigree): 17.7% weight loss in Chinese phase 3 at 48 weeks versus Zepbound’s 17.5% at ~52 weeks in comparable Chinese patients. Fazeli’s key nuance: Kailera’s curve “still seems to be continuing” where Zepbound plateaued at 44 weeks; Werber adds the adverse-event profile “definitively looks better,” dose escalation ate 5-6 months so patients are far from FDA’s 52-weeks-at-target-dose bar, an 8-mg dose is still being tested, and higher doses are planned, with a 4,500-patient phase 3 potentially starting by year-end — with a U.S. partner likely needed.
• Fazeli’s 53-physician U.S. lung cancer survey: AstraZeneca’s AVANZAR most anticipated as practice-changing (49%), Summit’s PD-1/VEGF ivonescimab ~40%, and the surprise — BMS’s relatlimab-plus-nivo lung combo cracked the list despite the field’s post-TIGIT/TIM-3 fatigue. BioNTech’s BNT327 scored poorly, which Sam attributes to only 5-7 U.S. trial centers: “familiarity is an issue.”
• DiaMedica’s preeclampsia protein therapeutic showed blood-pressure reduction and uterine-artery dilation in pregnant women — Schimmer doesn’t think the sector has seen a preeclampsia data update before, given how hard recruitment is when “parents want to do anything and everything… to make sure their baby is healthy.”
• Earnings: J&J reported strong results with conservative H2 guidance, cut its tariff impact from $400M to $200M, and Schimmer flagged Spravato up 30% quarter-over-quarter after reimbursement struggles. Novartis beat and nudged up operating-profit guidance but cut the 2025 Cosentyx forecast — 12% of 2024 sales, an $8B peak-sales estimate before a 2029 patent expiry, at least on paper — while announcing a $10B buyback; Fazeli: “they have enough money to keep doing M&A.”