Article

Sites unseen

Commercial due diligence, as the explainers teach it, sizes the market and counts the competitors. In a drug deal the value sits in a few public, checkable facts -- one trial site, one maintenance fee, one line of a label -- and in 2023 a buyer wired $700 million for a drug whose Phase 2 result an earlier bidder, in its own data-room review, had traced to a single site in Bulgaria.

commercial due diligence on a biotech or pharma asset -- what the generic checklist misses, which post-deal losses were readable before the wire, and what to check in 2026 · 2026-10-16 · 28 sources

Commercial due diligence is the work a buyer does before a deal to test whether a business is worth what the seller says: how big the market is, who the customers are, who else sells to them, and whether the plan survives contact with all three. Financial due diligence checks the books; legal checks the contracts; commercial checks the story. An unapproved drug tells a different story. There are no customers to interview and the market is years away. Its value sits in a short list of facts, most of them public and each checkable: whether the data behind the last readout holds up, whether the patents still stand, what the label will say, who will pay, whether the plant will pass inspection, and what the next readout must show. Biotech due diligence is the craft of reading those facts in the form that shows them. The generic checklist looks at none of them, so it cannot tell a priced bet that lost from a fact that sat unread in the data room.

The checklist was built for something else

We read the text of 15 ranking explainers of commercial due diligence, from Expert360 and Investopedia to the big consultancies and the data-room vendors, and searched them for the words that decide a drug deal. In the text we read, "clinical trial", "probability of success", "label", "patent", "FDA", "payer" and "Inflation Reduction Act" appear nowhere; "exclusivity" appears twice, in the commercial sense.1Our reading of 15 ranking explainers (Expert360 x2, Investopedia, eComplete x3, L.E.K., PwC, KPMG, Grant Thornton, Ansarada, Dealroom, SmartRoom, PrepLounge, Ellty), October 16, 2026, 5,000-12,000 characters per page; truncated, so the count is of the text read, not of the full pages. Also absent: "likelihood of approval", "CMC", "reimbursement". (Own measurement) Expert360's worked example is AOL-Time Warner; Investopedia's nine steps are for buying a share.1Our reading of 15 ranking explainers (Expert360 x2, Investopedia, eComplete x3, L.E.K., PwC, KPMG, Grant Thornton, Ansarada, Dealroom, SmartRoom, PrepLounge, Ellty), October 16, 2026, 5,000-12,000 characters per page; truncated, so the count is of the text read, not of the full pages. Also absent: "likelihood of approval", "CMC", "reimbursement". (Own measurement) None of it is wrong. It is about other businesses, in which the customer exists and can be telephoned.

The fair objection is that nobody runs a drug deal off Investopedia: L.E.K., Bain and ZS sell pharma-specific commercial diligence with probability-of-success models and payer work. True; and the reader who typed the phrase into a search box met the explainer. The practice adds what the explainer omits: a base rate. Citeline's Biomedtracker puts the industry's likelihood of approval for a drug entering the clinic at 6.7%.2pharmaphorum, "Clinical development rates are falling -- it's not all bad news", July 18, 2024, reporting Citeline Biomedtracker: 6.7% from Phase I (10.4% in 2014). pharmaphorum.com (T3 -- industry data, as reported) A buyer who prices from that number does something the generic checklist has no box for.

Three kinds of loss

Post-deal write-downs sort into three piles. Only one is diligence's fault.

The first is priced risk that lost. AbbVie's emraclidine, acquired with Cerevel, missed its primary endpoint in two Phase 2 trials in November 2024; the 8-K recorded "a non-cash after-tax intangible asset impairment charge of approximately $3.5 billion", with about $3.6 billion of Cerevel intangibles still on the books.3AbbVie Inc., Form 8-K Item 2.06, January 10, 2025: emraclidine missed the primary endpoint in two Phase 2 EMPOWER trials (announced November 11, 2024); remaining Cerevel intangibles "of approximately $3.6 billion". sec.gov (T1 -- SEC filing) Bristol Myers Squibb's Cobenfy, acquired with Karuna, missed statistical significance on the primary endpoint of its Phase 3 adjunctive trial in April 2025.4Bristol Myers Squibb, Form 8-K exhibit 99.1, April 24, 2025: the Phase 3 ARISE adjunctive trial of Cobenfy "did not meet the threshold for statistical significance for the primary endpoint". sec.gov (T1 -- SEC filing) Pfizer's 2024 accounts carry a $1.0 billion impairment on a Phase 1 asset that came with Seagen.5Pfizer Inc., Form 10-K for 2024, February 27, 2025: impairment of "$1.0 billion for B7H4V (felmetatug vedotin), a Phase 1 IPR&D asset", and of Tukysa and disitamab vedotin. sec.gov The 10-K does not name the source; ClinicalTrials.gov NCT05194072 lists the lead sponsor as "Seagen, a wholly owned subsidiary of Pfizer" (read October 16, 2026). (T1 -- SEC filing and registry record) In each case the buyer knew it was paying for a readout that had not happened; the loss was the price of the bet. Competition is a cousin of this pile, and the one checklist box that does real work in a drug deal: the same filing impairs $400 million on Tukysa and $200 million on disitamab vedotin "reflecting emerging competition".5Pfizer Inc., Form 10-K for 2024, February 27, 2025: impairment of "$1.0 billion for B7H4V (felmetatug vedotin), a Phase 1 IPR&D asset", and of Tukysa and disitamab vedotin. sec.gov The 10-K does not name the source; ClinicalTrials.gov NCT05194072 lists the lead sponsor as "Seagen, a wholly owned subsidiary of Pfizer" (read October 16, 2026). (T1 -- SEC filing and registry record)

The second pile is the event after close that no reading could have found. Two non-ambulatory patients died of acute liver failure after receiving Elevidys, Sarepta's gene therapy for Duchenne, which Roche holds the rights to outside America; Roche stopped commercial dosing of that group on June 15, 2025,6Roche, ad hoc announcement, June 15, 2025: commercial dosing of Elevidys in non-ambulatory patients stopped. roche.com (T1 -- the company's own disclosure) and the American label now carries a boxed warning and is limited to ambulatory patients aged four and over.7FDA label for Elevidys via openFDA drug/label, effective August 12, 2026: boxed warning for acute serious liver injury and acute liver failure; indication limited to ambulatory patients 4 years of age and older. (T1 -- regulator's record, read October 16, 2026) Pfizer withdrew Oxbryta from every market in September 2024.5Pfizer Inc., Form 10-K for 2024, February 27, 2025: impairment of "$1.0 billion for B7H4V (felmetatug vedotin), a Phase 1 IPR&D asset", and of Tukysa and disitamab vedotin. sec.gov The 10-K does not name the source; ClinicalTrials.gov NCT05194072 lists the lead sponsor as "Seagen, a wholly owned subsidiary of Pfizer" (read October 16, 2026). (T1 -- SEC filing and registry record) A label can shrink after approval; the revenue curve should carry that possibility, and no line in a data room would have shown it.

The third pile is the one the checklist cannot see and the buyer most needs to: facts that sat in the record or the data room, readable and unread. The sorting is not hindsight; it uses only what the buyer could see at signing.

The Bulgarian site

In February 2023 Novo Nordisk began diligence on ocedurenone, a KBP Biosciences drug for uncontrolled hypertension in chronic kidney disease, while its Phase 3 was running. Novo "had access to a full electronic data room". Among the documents KBP uploaded was a listing of the vital signs of every Phase 2 patient, "but in Portable Document Format".8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) The purchase agreement was signed on October 11, 2023 and closed on November 29; Novo wired $700 million and put $100 million in escrow.8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) On June 26, 2024 the Phase 3 interim analysis met its futility criteria, and Novo announced an impairment of more than $800 million.8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) The registry record: terminated, "Interim futility analysis showed primary endpoint not achieved", 652 enrolled.9ClinicalTrials.gov NCT04968184, ocedurenone Phase 3: status TERMINATED for interim futility, enrollment 652. (T1 -- registry record, read October 16, 2026)

Novo now alleges that the Phase 2 result rested on one site. The Singapore judgment records that an earlier bidder, Otsuka, concluded after its own data-room review of the Phase 2 results that a "Single site (Bulgaria) was responsible for bulk of efficacy, raising questions about repeatability", and did not proceed.8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) The Court of Appeal, setting out Novo's case as it dismissed KBP's appeal in May 2026, put it in one line: "Novo could have reached this conclusion by an analysis of the Vital Signs Data Listing, as indeed Otsuka had done".10KBP Biosciences Pte Ltd v Novo Nordisk A/S [2026] SGCA(I) 2, May 14, 2026, paras 21-25, 45, 91: appeal dismissed on all grounds. Para 22 summarises Novo's claims; para 24 lists the undisclosed Phase 2 interim analysis among the matters on which the judge found a good arguable case; para 25, ICC arbitration in New York, filed by Novo on March 26, 2025. The court decides only that a good arguable case exists. elitigation.sg (T1 -- court record) The Phase 2 clinical study report in the data room said that "No interim analysis for efficacy was planned for this study"; the same court lists an undisclosed Phase 2 interim analysis among the matters on which the judge found a good arguable case.10KBP Biosciences Pte Ltd v Novo Nordisk A/S [2026] SGCA(I) 2, May 14, 2026, paras 21-25, 45, 91: appeal dismissed on all grounds. Para 22 summarises Novo's claims; para 24 lists the undisclosed Phase 2 interim analysis among the matters on which the judge found a good arguable case; para 25, ICC arbitration in New York, filed by Novo on March 26, 2025. The court decides only that a good arguable case exists. elitigation.sg (T1 -- court record)

The legal posture is narrow and worth stating exactly. Both Singapore rulings decided only that Novo has a "good arguable case", the standard for the freezing order it obtained.11Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 3, February 14, 2025: freezing order up to US$730 million on the good-arguable-case standard. laws.sg (T1 -- court record) Whether KBP misrepresented anything is for an arbitration in New York, which has not reported, and KBP contests Novo's account.8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) Nothing here turns on who is right. Two bidders had the same patient listing. One read efficacy by site and walked away; the other did not. The first-instance judge observed that Novo "would not have known where to look for the anomaly if it was not alerted".8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) That is a sentence about a data room. The test is whether anyone reads the data in the form that shows the problem; a per-site efficacy table from a patient listing is an afternoon's work. The public registry would not have helped: the Phase 2b record lists two locations, both in the United States, while the judgment says the trial recruited in seven countries, Bulgaria among them.12ClinicalTrials.gov NCT03574363, ocedurenone Phase 2b: two locations listed, both in the United States; enrollment 162; results posted. Whether other sites ran under another record or were never registered is not established. (T1 -- registry record, read October 16, 2026)

The objection writes itself: diligence cannot beat a seller who hides things, if that is what happened. Partly conceded. The warranty did the legal work, and the escrow cap of $100 million absent fraud measures what the contract alone would have returned.8Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) But the anomaly was readable without any allegation at all, and a bidder read it.

Exclusivity is a stack, and sometimes a lapsed fee

"Patented until 20XX" is the slide the commercial model ends on, since the revenue curve stops where exclusivity stops, and it is a claim to check at the registry rather than accept from the deck. An American patent requires maintenance fees at three and a half, seven and a half and eleven and a half years; unpaid, it expires, and nobody writes to tell the buyer.1335 U.S.C. 41(b) and 37 CFR 1.362: maintenance fees due at 3 years 6 months, 7 years 6 months and 11 years 6 months from grant; the patent expires on non-payment, after the grace period. (T1 -- statute and regulation) In a seed deck we read for an angel group's screening committee in May 2026, the "2 Patents Issued" slide named two patents whose status at the patent office reads "Patent Expired Due to NonPayment of Maintenance Fees": one lapsed at the 7.5-year fee, the other at the 11.5-year fee, months before the deck reached the committee.14USPTO Open Data Portal transaction histories, read October 16, 2026: expiry recorded June 2, 2025 (7.5-year fee) and January 26, 2026 (11.5-year fee; eighth-year fee paid June 8, 2021). Deck to the committee May 5, 2026. Company and patent numbers withheld, as in "Pipe dreams"; the delivery is in data/career_history.md. (T1 -- patent office record) A stale slide rather than a deception, and the finding went to the committee as a question; the full walk is in Pipe dreams.

The error runs the other way too: Keytruda's key patents expire in 2028, but Bloomberg Intelligence argues its layered portfolio holds off erosion until about 2033.15BioSpace, February 24, 2026, citing Bloomberg Intelligence: the layered portfolio is worth about $22 billion of sales. biospace.com (T4 -- news report of an analyst estimate) And a second clock now runs beside the patent clock: under the Inflation Reduction Act a small molecule becomes eligible for Medicare price negotiation nine years after approval, a biologic after 13.16Vogel M, et al. JAMA Health Forum, October 17, 2025, doi:10.1001/jamahealthforum.2025.4720: Medicare negotiation eligibility under the Inflation Reduction Act. (T2 -- peer-reviewed) Exclusivity is a stack of patents, regulatory grants and statutory clocks, each with its own date, and the model ends on the shortest. Patent diligence is the lawyers' job, and it arrives late; the registry check takes minutes and the revenue curve depends on it, so the commercial read runs it first.

Most of the headline is contingent

By most counts 2025 was a record year for drug deals, and the counts disagree by definition: $133 billion over 50 majority-stake acquisitions of at least $250 million by IQVIA's rule,17IQVIA, "Biopharma M&A outlook for 2026", January 19, 2026. iqvia.com (T3 -- industry report) $138 billion over 129 deals by J.P. Morgan's.18J.P. Morgan, Q4 2025 Biopharma Licensing and Venture Report. jpmorgan.com (T3 -- bank research) Licensing passed $252 billion, the highest since 2016, with upfronts "steady at 7% of total deal value".18J.P. Morgan, Q4 2025 Biopharma Licensing and Venture Report. jpmorgan.com (T3 -- bank research) China supplied $135.655 billion of out-licensing across 157 deals and collected about $7 billion upfront, which is 5.2% by our arithmetic.19Han Kun Law Offices, January 13, 2026: China outbound innovative-drug business development. The 5.2% upfront share is our computation (7 / 135.655). hankunlaw.com (T3 -- law-firm report) Contingent value rights sat in under 10% of public biopharma acquisitions before 2019 and in 23-33% every year since 2023; net-sales milestones appeared in five deals, all in 2025-26.20Gibson Dunn, "How the most-favored-nation regime affects biopharma deals", August 31, 2026. gibsondunn.com (T3 -- law-firm analysis)

The structural argument says that milestones price the risk, so diligence matters less. They shift the risk; they leave the upfront, the opportunity cost and the competitor who took the asset instead. And a milestone priced on a Phase 2 that rested on one site is mispriced whatever its structure. Licensing due diligence asks the same four questions as an acquisition but sharpens the readout question, because the next readout is usually what the milestone is written on.

The live example is Summit's ivonescimab, licensed from Akeso. In HARMONi the overall-survival hazard ratio was 0.79 (95% CI 0.62-1.01) at the primary analysis in April 2025, short of significance; 0.78 (0.62-0.98; nominal p=0.0332) in September 2025; and 0.76 in June 2026, both overall and in the Western subgroup, whose follow-up stood at 23.2 months against 32.7 for the Asian patients.21Summit Therapeutics, Form 8-K exhibit 99.1, July 22, 2026: HARMONi overall survival, ITT and Western populations; PDUFA goal date. sec.gov (T1 -- SEC filing) The FDA's goal date is November 14, 2026, a test of the data on which this article makes no call.21Summit Therapeutics, Form 8-K exhibit 99.1, July 22, 2026: HARMONi overall survival, ITT and Western populations; PDUFA goal date. sec.gov (T1 -- SEC filing) The diligence point is the moving confidence interval: before any wire, settle which cut of a result the price was written on.

What to check in 2026

The list got longer this year. Six checks, in the order a buyer meets them.

Data, per site. Ask for the patient listing, tabulate efficacy by site and by region, and compare the registry record with the clinical study report.12ClinicalTrials.gov NCT03574363, ocedurenone Phase 2b: two locations listed, both in the United States; enrollment 162; results posted. Whether other sites ran under another record or were never registered is not established. (T1 -- registry record, read October 16, 2026)

Inspectability. In September 2026 the FDA's centre directors wrote that the agency "will be increasing its attention to clinical studies (and sites) not conducted under IND", that it may refuse data from sites it cannot inspect, and that sponsors should treat the inability to inspect "as a material factor in their regulatory strategy, not an administrative footnote".22FDA Voices, "Good Clinical Practices Are Not Optional", by the directors of CDER, CBER, CDRH and the Oncology Center of Excellence. The page is undated; September 2, 2026 is the date of the trade reports on it. fda.gov (T1 -- regulator's statement)

Patents and clocks. Registry status and fee history for every number on the slide; regulatory exclusivity; the negotiation clock and orphan status.

Label and payer. Under the most-favoured-nation policy an American price is tied to the lowest in comparable developed countries, so a partner's launch abroad can set the reference at home;23Skadden, "MFN drug pricing and its impact", June 4, 2026: ex-US licences implicated. skadden.com (T3 -- law-firm analysis) lawyers report clients asking them to delay launches outside America to avoid it.24Endpoints News, March 17, 2026: lawyers fielding requests to delay ex-US launches to avoid setting a reference price. endpoints.news (T4 -- news report)

Manufacturing. The FDA now publishes its complete response letters, 458 of them on openFDA at the last update.25openFDA transparency/crl endpoint: 458 complete response letters, last updated August 13, 2026. api.fda.gov (T1 -- regulator's archive, read October 16, 2026) In a peer-reviewed sample of 43 new molecular entities that received one, facility deficiencies appeared in 65%, chemistry, manufacturing and controls in 51%, and efficacy in 26%.26Dilek S, et al. Therapeutic Innovation and Regulatory Science, February 11, 2026, PMID 41670800, doi:10.1007/s43441-026-00921-3; the same sample: labeling 44%, safety 26%. (T2 -- peer-reviewed) The plant fails more often than the drug. The BIOSECURE Act, enacted on December 18, 2025 in the defence authorisation bill, asks who the contract manufacturer is.27Arnold & Porter, "The BIOSECURE Act becomes law in the United States", December 23, 2025: section 851 of the FY2026 National Defense Authorization Act, P.L. 119-60. arnoldporter.com (T3 -- law-firm advisory; statute text not read)

The next readout. Write down, dated, what result makes the deal work, so the price can be scored against it; What a verified read looks like applies that method to one 8-K.

An outside read cannot replace a key-opinion-leader call or a site audit. It surfaces where the primary record and the pitch disagree; whether a gap is fatal or fixable stays with the buyer and the committee. We have sat on an angel group's diligence teams since 2015, about 25 deals by our own estimate, and in 2026 we ran the method on two seed decks: received May 5, both reads delivered May 11, about 20 claims per report checked by hand, the expired patents above the lead finding.28BigBio house record, data/career_history.md: Mass Medical Angels due-diligence teams since 2015; "I fact checked ~20 claims per report by hand". Unpaid pilot. (T1 -- first-party record) That is the read we sell, and it starts with one question: where does the record disagree with the deck?

Footnotes

  1. Our reading of 15 ranking explainers (Expert360 x2, Investopedia, eComplete x3, L.E.K., PwC, KPMG, Grant Thornton, Ansarada, Dealroom, SmartRoom, PrepLounge, Ellty), October 16, 2026, 5,000-12,000 characters per page; truncated, so the count is of the text read, not of the full pages. Also absent: "likelihood of approval", "CMC", "reimbursement". (Own measurement) ↩ ↩2

  2. pharmaphorum, "Clinical development rates are falling -- it's not all bad news", July 18, 2024, reporting Citeline Biomedtracker: 6.7% from Phase I (10.4% in 2014). pharmaphorum.com (T3 -- industry data, as reported) ↩

  3. AbbVie Inc., Form 8-K Item 2.06, January 10, 2025: emraclidine missed the primary endpoint in two Phase 2 EMPOWER trials (announced November 11, 2024); remaining Cerevel intangibles "of approximately $3.6 billion". sec.gov (T1 -- SEC filing) ↩

  4. Bristol Myers Squibb, Form 8-K exhibit 99.1, April 24, 2025: the Phase 3 ARISE adjunctive trial of Cobenfy "did not meet the threshold for statistical significance for the primary endpoint". sec.gov (T1 -- SEC filing) ↩

  5. Pfizer Inc., Form 10-K for 2024, February 27, 2025: impairment of "$1.0 billion for B7H4V (felmetatug vedotin), a Phase 1 IPR&D asset", and of Tukysa and disitamab vedotin. sec.gov The 10-K does not name the source; ClinicalTrials.gov NCT05194072 lists the lead sponsor as "Seagen, a wholly owned subsidiary of Pfizer" (read October 16, 2026). (T1 -- SEC filing and registry record) ↩ ↩2 ↩3

  6. Roche, ad hoc announcement, June 15, 2025: commercial dosing of Elevidys in non-ambulatory patients stopped. roche.com (T1 -- the company's own disclosure) ↩

  7. FDA label for Elevidys via openFDA drug/label, effective August 12, 2026: boxed warning for acute serious liver injury and acute liver failure; indication limited to ambulatory patients 4 years of age and older. (T1 -- regulator's record, read October 16, 2026) ↩

  8. Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 22, August 12, 2025, paras 4, 14-19, 41, 43-45, 71; Phase 2 recruited in the US, Chile, Georgia, Bulgaria, Hungary, Israel and Australia. The ruling finds only a good arguable case; the merits are for the New York arbitration. Reuters (reuters.com, February 18, 2025) calls it a "$1.3 billion deal"; the judgment's figures are used here. elitigation.sg (T1 -- court record) ↩ ↩2 ↩3 ↩4 ↩5 ↩6 ↩7

  9. ClinicalTrials.gov NCT04968184, ocedurenone Phase 3: status TERMINATED for interim futility, enrollment 652. (T1 -- registry record, read October 16, 2026) ↩

  10. KBP Biosciences Pte Ltd v Novo Nordisk A/S [2026] SGCA(I) 2, May 14, 2026, paras 21-25, 45, 91: appeal dismissed on all grounds. Para 22 summarises Novo's claims; para 24 lists the undisclosed Phase 2 interim analysis among the matters on which the judge found a good arguable case; para 25, ICC arbitration in New York, filed by Novo on March 26, 2025. The court decides only that a good arguable case exists. elitigation.sg (T1 -- court record) ↩ ↩2

  11. Novo Nordisk A/S v KBP Biosciences Pte Ltd [2025] SGHC(I) 3, February 14, 2025: freezing order up to US$730 million on the good-arguable-case standard. laws.sg (T1 -- court record) ↩

  12. ClinicalTrials.gov NCT03574363, ocedurenone Phase 2b: two locations listed, both in the United States; enrollment 162; results posted. Whether other sites ran under another record or were never registered is not established. (T1 -- registry record, read October 16, 2026) ↩ ↩2

  13. 35 U.S.C. 41(b) and 37 CFR 1.362: maintenance fees due at 3 years 6 months, 7 years 6 months and 11 years 6 months from grant; the patent expires on non-payment, after the grace period. (T1 -- statute and regulation) ↩

  14. USPTO Open Data Portal transaction histories, read October 16, 2026: expiry recorded June 2, 2025 (7.5-year fee) and January 26, 2026 (11.5-year fee; eighth-year fee paid June 8, 2021). Deck to the committee May 5, 2026. Company and patent numbers withheld, as in "Pipe dreams"; the delivery is in data/career_history.md. (T1 -- patent office record) ↩

  15. BioSpace, February 24, 2026, citing Bloomberg Intelligence: the layered portfolio is worth about $22 billion of sales. biospace.com (T4 -- news report of an analyst estimate) ↩

  16. Vogel M, et al. JAMA Health Forum, October 17, 2025, doi:10.1001/jamahealthforum.2025.4720: Medicare negotiation eligibility under the Inflation Reduction Act. (T2 -- peer-reviewed) ↩

  17. IQVIA, "Biopharma M&A outlook for 2026", January 19, 2026. iqvia.com (T3 -- industry report) ↩

  18. J.P. Morgan, Q4 2025 Biopharma Licensing and Venture Report. jpmorgan.com (T3 -- bank research) ↩ ↩2

  19. Han Kun Law Offices, January 13, 2026: China outbound innovative-drug business development. The 5.2% upfront share is our computation (7 / 135.655). hankunlaw.com (T3 -- law-firm report) ↩

  20. Gibson Dunn, "How the most-favored-nation regime affects biopharma deals", August 31, 2026. gibsondunn.com (T3 -- law-firm analysis) ↩

  21. Summit Therapeutics, Form 8-K exhibit 99.1, July 22, 2026: HARMONi overall survival, ITT and Western populations; PDUFA goal date. sec.gov (T1 -- SEC filing) ↩ ↩2

  22. FDA Voices, "Good Clinical Practices Are Not Optional", by the directors of CDER, CBER, CDRH and the Oncology Center of Excellence. The page is undated; September 2, 2026 is the date of the trade reports on it. fda.gov (T1 -- regulator's statement) ↩

  23. Skadden, "MFN drug pricing and its impact", June 4, 2026: ex-US licences implicated. skadden.com (T3 -- law-firm analysis) ↩

  24. Endpoints News, March 17, 2026: lawyers fielding requests to delay ex-US launches to avoid setting a reference price. endpoints.news (T4 -- news report) ↩

  25. openFDA transparency/crl endpoint: 458 complete response letters, last updated August 13, 2026. api.fda.gov (T1 -- regulator's archive, read October 16, 2026) ↩

  26. Dilek S, et al. Therapeutic Innovation and Regulatory Science, February 11, 2026, PMID 41670800, doi:10.1007/s43441-026-00921-3; the same sample: labeling 44%, safety 26%. (T2 -- peer-reviewed) ↩

  27. Arnold & Porter, "The BIOSECURE Act becomes law in the United States", December 23, 2025: section 851 of the FY2026 National Defense Authorization Act, P.L. 119-60. arnoldporter.com (T3 -- law-firm advisory; statute text not read) ↩

  28. BigBio house record, data/career_history.md: Mass Medical Angels due-diligence teams since 2015; "I fact checked ~20 claims per report by hand". Unpaid pilot. (T1 -- first-party record) ↩