Episode 189 - July 17, 2026

Summary

• Lilly’s “Amazonification” continued with a $3.8B deal for atai Beckley — $2.8B upfront plus a $1B CVR — buying a nasally administered depression treatment ahead of its pivotal readout. Maraganore called it “a strong and aggressive bet” but arguably the right one given conviction, and noted it was one of the first deals under Carol Ho’s leadership of Lilly neuroscience. He tied it to the Economist thesis that Dave Ricks is remaking the industry’s first trillion-dollar business into “something closer to a tech company” — prevention over treatment, with LillyDirect’s telehealth/DTC model effectively going around PBMs. Fazeli’s read: Lilly is pre-empting the potential “plateau or pressure” on its obesity franchise a decade out.
• BioCentury’s report that roughly 40% of 14–15 new drugs originating from China are now first-in-class drew a contrarian frame from Gline: China is “a little bit of a smoke screen or a red herring” for the commodification of antibody and small-molecule discovery. Close it off and “it would just happen somewhere else”; America’s enduring edge is that everyone wants U.S. late-stage trials and approval — “a lot of that has to do with our high pricing.” On data quality he’s pragmatic: speed and variability will trade off, but “the proof’s in the pudding” — the Phase 3 goes on the label.
• A report that FDA advised a Chinese company not to file because its drug is manufactured in China — despite data being generated in a U.S.-conducted trial — was flagged by Fazeli as double-checked but unexplained. Gline’s bounds: dozens of approved U.S. drugs are made in part in China, so this is either one site not up to snuff, or “a strong political commentary by political appointees at HHS that could have drastic implications.” Something to watch, not yet a theme.
• Roivant’s brepocitinib launch in dermatomyositis is imminent — Werber models a $6–12B product across four pivotal indications — and Gline placed pricing between IVIG at ~$225K and Vyvgart at ~$870K gross, “closer to the lower end.” He won’t guide launch numbers (“y’all are mean to companies that do”) but stressed the orphan playbook: small budget impact, not rebate-intensive, bridge programs, and lessons from Alnylam-style launches — “stand on the shoulders of giants.”
• BIO sharply rebuked the OMB proposal on political influence over federal grants, while Kalshi announced a venue for betting on late-stage trials and approvals, with insider guardrails and companies around a $500M market cap. It already prices brepocitinib at 81% approval by year-end, with about $3,000 staked per side. Roivant is debating whether its insider-trading policy covers it; Gline is skeptical anyone validates the tool soon — “it would take a lot of n” — while Fazeli wants it tested as a new probability-of-success input. Werber’s stickler is regulatory: physicians or nurses betting on open-label studies they’re in.
• Merck’s oral PCSK9, enlicitide, was approved at $3,800/year — under the roughly $5–6K antibodies and mRNA products — under Marty’s CNPV program, and Maraganore predicts it does well. His Corsera counter-thesis: a once-yearly PCSK9 plus angiotensinogen siRNA for primordial prevention, approvable on biomarkers with an AI-enabled lifetime-risk tool as “the key” to the approach. Per Fazeli, if that works and is priced affordably, why would AstraZeneca’s oral obesity-pill combination be needed?
• Biogen/Ionis’s intrathecal tau ASO technically failed its Phase 2 primary on an inverted dose response, but Maraganore is “net encouraged” on tau — blaming ASO tolerability, not the target, and citing Roche’s tominersen termination and eplontersen’s CARDIO-TTRansform failure as reminders of the complicated ASO story. The reported slowing on ADAS-Cog and CDR-SB was 0.54 points versus the FDA’s 0.5 bogey. Werber defended trying biology at the edge of understanding — “somebody has to do this” — while Maraganore, whose mother has Alzheimer’s, called the absence of an ARIA signal a “big deal” versus the antibodies’ monitoring burden.
• Rapid-fire: AstraZeneca licensed Dizal’s EGFR exon-20 inhibitor for $600M upfront plus up to $900M in milestones — Fazeli says the data “stacks up” versus Rybrevant and other competitors, and the Tagrisso 800-pound gorilla is the ideal owner — while Erasca’s RAS data improved with more patients and follow-up and its raise was upsized to about $600M despite Revolution Medicines IP questions.

Deep dive

1. Even the CEO of the company can’t read his own tape

• Gline’s Friday anecdote, told as a checklist of dead ends: stock down 4–5%, first thought “shoot, what did I say?”, then a competitor, then an MFN-pricing headline, then the Fed — all wrong. The bank’s answer: “It’s factor driven… It’s Jane Street. Blame Jane Street.” His friend at Jane Street: “It’s not us.” The meta-lesson from someone who says his team probably knows Roivant’s trading dynamics better than anyone: “I think it’s just impossible to know some days, and I found that like an almost zen thought.”
• Macro isn’t a major focus for Roivant — “we’ve been on a good run and have a lot to do” — though Gline watches whether the market dynamic around companies like his changes, with “read-through effects to strategy.”

2. Lilly is pre-hedging its own obesity peak

• The deal: atai Beckley, $3.8B — $2.8B upfront, $1B CVR — for a nasally administered treatment for depression in pivotal trials. Buying ahead of the readout is, per Maraganore, “a strong and aggressive bet… but arguably the right way to do it if they have conviction in the mechanism.” He also flags it as one of the first deals under Carol Ho, the physician-scientist Lilly poached from Denali, where she was chief medical officer, to lead neuroscience.
• The Economist’s “Lilly reinventing the pharma business” piece got Maraganore’s endorsement: the industry’s first trillion-dollar business shifting from treating disease to preventing it, borrowing Silicon Valley playbooks, and using LillyDirect’s telehealth/DTC model to effectively go around the PBMs — “closer to a tech company at some level.”
• Fazeli owns the “Amazonification” coinage: three infectious-disease deals almost back-to-back a couple months ago, now a third CNS area outside neurodegeneration — probably the only way to manage the eventual “plateau or pressure” on the obesity franchise ten years out. Maraganore’s addendum: most companies with a decade of growth would “let it ride”; Lilly is showing “nothing less than incredible urgency.”

3. China is the venue, not the story — but watch that manufacturing advisory

• BioCentury’s report: roughly 4 of 14–15 new drugs originating from China, or about 36–40%, are now first-in-class. Its Asia-to-West NewCo framing included a couple of Japanese assets. Gline’s standing thesis: China is “a smoke screen or a red herring” for the commodification of antibody development and small-molecule chemistry, with or without AI — cut off China and “it would just happen somewhere else.” The U.S. retains the prize because everyone wants U.S. late-stage trials and approval, and “a lot of that has to do with our high pricing” — so global innovation ends up benefiting American patients.
• On data trust, Werber and Maraganore revisited a debunked morning-vs-evening Keytruda dosing study. Maraganore said “it is a bit of a wild west… buyer beware,” with China’s speed cutting both ways. Gline granted real speed/quality trade-offs: “the proof’s in the pudding… the Phase 3 study is going to be the answer that goes on the label.”
• The tuck-in: Fazeli double-checked a report that FDA advised a company not to file because its product is manufactured in China, even though the data were being generated in a U.S.-conducted trial. Gline’s bracket: dozens of approved drugs are partly made in China, so it’s either a site that “wasn’t up to snuff” being spun as an anti-China stance, or “a strong political commentary by political appointees at HHS that could have drastic implications.”

4. Brepocitinib pricing: bookends confirmed, “closer to the lower end”

• Werber’s setup: PDUFA and September-ish launch in dermatomyositis for Roivant’s oral JAK1/TYK2 after positive VALOR Phase 3; he models $6–12B across four pivotal indications, with noninfectious uveitis Phase 3 data by year-end; 20–40% of patients already on off-label JAKs, with physicians warehousing patients who want to switch; standard-of-care IVIG runs ~$225K and is “extremely inconvenient.”
• Gline refused specific launch guidance — “y’all are mean to companies that do” — but treated the bookends as roughly IVIG at $225K and Vyvgart at ~$870K gross with weekly dosing, while noting that prior Vyvgart discussions used a $500–600K figure. He expects brepocitinib “closer to the lower end of that range than the top end.” And a preemptive wink: “to all the payers listening… this is not going to be a huge budget impact.”
• The launch mechanics per Gline: orphan-ish, high-morbidity markets aren’t rebate-intensive — “it’s not like there’s four big pharma companies competing with rebates in dermatomyositis” — so it’s bridge programs, patient support, and payer navigation. He’s studying successful orphan launches including Alnylam’s to “stand on the shoulders of giants”; Maraganore adds that proactive value-based agreements at the beginning of launches, enabled by outstanding data and manageable budget impact, reduced ordinary headwinds.

5. Policy and the trial casino: BIO rebukes OMB; Kalshi lists brepocitinib at 81%

• Maraganore welcomed BIO’s “pretty sharp rebuke” of the OMB proposal to let political operatives decide whether to continue or authorize federal grant funding: it would destabilize the research ecosystem, weaken merit-based funding, and undermine U.S. competitiveness and health security. His broader point: “we all need to be vocal on… administration policies — any administration, not just this one — that can weaken innovation.”
• Kalshi announced a venue for betting on late-stage trial outcomes and approvals, with insider guardrails and participation restricted to companies with market caps of about $500M. Fazeli, disclosing Bloomberg’s partnership with AppliedXL, takes the academic line: let it run unless regulators stop it, and eventually test whether the markets predict outcomes — “a new probability of success into our calculations.” Gline’s cold water: “it would take a lot of n,” and in 20 years it’ll be “Claude betting against OpenAI… we’re just going to watch the robots talk to each other.”
• The live example: Kalshi gives brepocitinib an 81% chance of approval by year-end, with the order book described as around 81 at 74 and ~$3,000 staked per side. Roivant debated internally whether its insider-trading policy forbids employees from participating — “I would not be happy to find out that my employees were spending time trading.” Werber’s sticking point is regulatory: physicians and nurses betting on big open-label studies “would not be good.”

6. Oral PCSK9 arrives at $3,800 — and the once-a-year counterargument

• Merck’s enlicitide was approved as the first oral PCSK9 drug at $3,800/year versus roughly $5–6K for the antibodies and mRNA products already on the market, under Marty’s CNPV program. Maraganore: “I’m going to predict that it will do well… let’s see if Kalshi agrees.” AstraZeneca is following with a macrocyclic peptide, and its stated ambition includes combining its oral PCSK9 with its oral obesity drug — to which Fazeli responds: if John’s annual drug is “priced more affordably,” why do you need that?
• Maraganore’s Corsera path: PCSK9 plus angiotensinogen siRNA dosed once a year for primordial prevention, with two approval studies — one biomarker-based, one outcomes-based — since outcomes “have not been required for approval for a while” but support the business case. The outcomes trial must be much larger given low event rates in a population that has not really been studied, with enrichment making it more tractable; the AI-enabled lifetime-risk prediction tool is “the key” to the approach. His adherence argument: once-yearly dosing could reduce the disutility versus a daily pill, where statin-based adherence is “really prominent and problematic.”

7. Tau ASO: a failed primary that didn’t damn the target

• The data as Werber laid it out: Biogen/Ionis’s intrathecal tau ASO missed its 76-week primary (high dose vs. placebo) with an inverted dose response — the low dose looked better, consistent with Phase 1b — while reducing phospho-tau and tau imaging and producing a reported 0.54-point slowing on ADAS-Cog and CDR-SB against the FDA’s 0.5 clinical-meaningfulness bogey. Biogen is going to Phase 3 anyway; the stock fell, with the Street noting efficacy no better than Leqembi and Kisunla.
• Maraganore is “overall net encouraged… as it relates to tau”: the inverted dose response looks like ASO tolerability — the high dose had more AEs and lower dose density — and the broader ASO story has also included Roche terminating tominersen in Huntington’s and eplontersen failing in CARDIO-TTRansform. “I don’t think we should damn the target just because of the ASO.” He’s watching Arrowhead’s transferrin-receptor-shuttle tau program and Alnylam’s intrathecal one.
• Werber asked about the on-target risk of reducing total tau, given that the protein has a function; Fazeli said the issue would be on-target rather than off-target pharmacology at higher doses, but that the published biomarker data do not support it. The dose-response error bars for tau reduction overlap.
• Werber defended Biogen’s effort: “Biogen pays the price for trying something at the edge of our understanding of biology… somebody has to do this and figure it out.”
• Maraganore’s personal stake — his mother has Alzheimer’s — sharpened the practical case: the antibodies were “a cork in a fire hydrant,” ARIA monitoring is difficult “even if you live in New York City,” and patients don’t want to be reminded of the disease over and over. The lack of an ARIA signal so far is a “big deal”: if there’s only potential upside, “it changes the equation a little bit.”

8. M&A rapid fire: AstraZeneca–Dizal stacks up; Erasca’s raise is the tell

• Fazeli on AstraZeneca licensing Dizal’s EGFR exon-20 inhibitor sunvozertinib — $600M upfront, up to $900M in development milestones, on top of an existing investment: “we were wondering why it took them so long.” The data stack up against Cullinan, Takeda, Hansoh and J&J’s Rybrevant, with an easier side-effect profile, and the Tagrisso owner, with sales over $7B, is “the ideal partner” for a small indication that needs drugs.
• On Erasca: RAS data improved with four more patients and a month more follow-up — “we’re used to the opposite” — with ORR beginning to reach the range seen from Revolution Medicines, amid questions about possible Revolution IP infringement. The upsized ~$600M raise in a dodgy tape “sounds like somebody believes in this,” with a run of pancreatic-cancer catalysts ahead.