ABCT 103 · L6 · Strategic IP Management for Biotech Start-ups

📚 Lecture 6 · Strategic IP Management for Biotech Start-ups

The founder’s checklist: what to file, where, and when — learned through four mistake-clusters.

103 · L6 Strategic IP Management for Biotech Start-ups

Overview — the master logic of strategic IP

L1–L5 built up the legal machinery: what can be protected, how it is protected, how it is attacked. L6 answers the question a founder actually has to answer: given everything that machinery, what should I file, where, and when? The through-line is that IP strategy is not a legal checklist but a set of coordinated business decisions — every choice about timing, jurisdiction and ownership is downstream of a commercial objective, and every one of them interacts with the regulatory timetable. The lecture is built as a sequence of common mistakes, each illustrated by a real case: publishing too early, filing without a strategy, filing in the wrong jurisdiction first, and getting ownership wrong.

The one sentence to remember Never publicly disclose until you file, file early with a strategy rather than a reflex, remember that "filing a patent" is not the same as "owning a granted, enforceable patent", and treat healthcare and biopharma IP as a specialist problem — because a very common, very expensive error is to treat it as a general one.

The six questions of a strategic IP portfolio

The framework is organised as a set of questions you must answer before filing anything. Notice that only two of them are legal questions; the rest are commercial, geopolitical and temporal.

Question	What it decides
WHAT is your business objective for IP?	防禦或進攻——決定整個 portfolio 的性格
WHERE was the invention made?	是否觸發中國的對外申請許可（foreign filing licence）
WHERE are the inventors' residence / citizenship?	第一位申請人國籍、住所影響PCT 可否受理
WHERE will ownership of the IP sit?	控股架構、CRO 關係、前僱主權利
WHERE do you need protection?	製造（make）／使用（use）／銷售（sell）三地
WHEN are key events happening?	申請時序要配合論文、發表、路演、盡調、IPO

WHAT — defensive or offensive

The first question is not legal at all: it is a statement of intent. IP is either there to prevent others suing you, or there to enforce against competitors. Almost every practical decision downstream — how broad to draft, whether to disclose a trade secret instead of patenting, whether to license in or license out — follows from this one choice.

Posture	Objective	Typical instruments
Defensive	Prevent others from suing you	Patent（阻嚇作用）；trade secret（若對方無法逆向）
Offensive	Enforce against competitors	Patent + licensing；一旦對方逆向無法迴避，專利即成武器

WHY get a patent? — the reasons, ranked by how often they are the real one

Read the reasons in order The slide lists the motivations roughly in ascending order of seriousness. The top two are vanity (“looks good on my resume”, “patent pending” on a college application or a pitch deck) and financing optics; the bottom three are the reasons that actually pay for the whole programme.

Looks good — to write “patent pending” on your college application or product page. Cosmetic.
Establish what I invented — to operate before I start collaborating with another person. This one is real: it defines the boundary of what you bring into a collaboration.
Ensure freedom to operate before collaborating — know you are not stepping on someone else's patent. This one is real and frequently skipped.
Stop others from copying — only works if the invention cannot be reverse-engineered around. A patent on a readily-reversible molecule is a weak deterrent.
Get licensing fees — IP is an asset that can grow in value over time.
Raise money from investors / worthy investment — IP is a balance-sheet asset that supports a valuation.

WHERE was the invention made, and who owns it

Where was the invention made? Some countries have export restrictions on IP; some jurisdictions are not part of the PCT at all — Taiwan is the example given in this lecture, so a PCT route does not cover it and a direct national filing is required.
Inventors' residence and citizenship determine who qualifies to file first and whether a PCT application can be received.
Where will ownership sit? This is a corporate-structuring question, and in China it interacts with the security-clearance regime (see L6's Jiecang case).

WHERE do you need protection? — make, use, sell

Protection follows the commercial footprint, not the invention. Three places matter, and they are frequently different countries.

Footprint	Activity	Who you are protecting against
Make	Manufacturing	製造地的仿製者
Use	R&D	同領域的 copycats 與競爭者
Sell	Major markets	市場上的侵權者

The factors that decide where you actually file:

Factor	Why it changes the decision
Strength of protection	同一件發明在某國可專利、在另一國不可（見 L5 的自然物適格性）
Types of protection	能否用 trade secret 或版權代替專利
Scope of protection	保護範圍與製造／使用／銷售的關係
Size of market	小市場不值得單獨維護
Likelihood of copying	若難以逆向，專利的嚇阻力本身就低
Location of competitors	對手在哪裡，就在哪裡布防

WHEN — aligning filings with business events

The timeline slide is the operational heart of the lecture: two filing generations, a PCT, direct filings, and four business events that must be sequenced against them.

Generation	Filing	Triggered by / aimed at
Gen 1	File 1st provisional application	守住最早優先權；必須在任何公開之前
Gen 2	File 2nd provisional application	新增發明點（此處必須有新內容，不能只是重複）
International	File PCT	自優先權日起 12 個月內，決定在各國的佈防
National	File direct in China or US	PCT 未涵蓋、或需要當地專利才能取得方法論保護的法域

The business events that constrain the timeline, in the order they occur:

Publication in a scientific journal — the point of no return for novelty.
Product demo at a conference — public disclosure risk.
Newspaper article about the product.
Due diligence with a potential investor — a patent portfolio is priced in; an unfiled invention is valued at zero.
Product launch.
GRANT — the only event at which you actually own an enforceable right.

Exam focus The correct planning sentence from the slide: think about your purpose in filing; determine your priorities over timing of any public disclosure, scope and strength of protection needed, timeline of R&D and commercial efforts, and budget/cost; then choose the best long-term strategy weighing risks and benefits according to business goals. Every element of that sentence is examinable as a separate item.

The three timelines must interlock

The strategy slide is deliberately a Venn diagram with Timelines are linked in a number of ways at the centre:

BUSINESS PLAN
PATENT TIMELINE / DEADLINES
REGULATORY TIMELINE

And four rules follow: must consider a global protection strategy; patent + regulatory + market strategy go together; consider IP strategy as a whole (patents, regulatory exclusivities, trade secrets, trademarks); and be prepared for changes.

Intersection of IP and regulatory: patent linkage, data and market exclusivity

The lecture deliberately revisits the linkage mechanics from L5, because IP strategy and regulatory strategy are not two plans — they are one plan expressed in two vocabularies.

The Orange Book, restated for China Creation of the “Orange Book” — eligible entries are small molecules, biologics, traditional Chinese medicine, composition of matter, formulations, medical use, and extraction methods (TCM) with an NDA on file. Process: generics must make a certification for all “Orange Book” listed patents before they can get marketing approval (MA). Timeliness: if a generic declares a listed patent invalid or not infringed, the patent owner must initiate contentious action within 45 days; the 30-month stay is struck through and the 9-month stay for small molecule litigation applies; the outcome determines whether the generic gets MA. Reward: the first generic gets 12-month market exclusivity, plus 6 months' data exclusivity (underlined).

Category	聲明內容	Approval timeline
I	沒有任何與原研藥相關的專利列於名單	立即
II	名單上的專利已過期或已被無效	立即
III	有列名專利，但該藥不落入專利範圍	立即，但須等到專利到期才可上市
IV	申請人認為列名專利無效或未被侵權	創新公司須在聲明公告後 45 日內起訴或向 CNIPA 提出無效宣告；9 個月停留期

And what follows the litigation or the opposition:

結果	審批時間表
仿製藥不落入專利／專利被無效／雙方和解	正常行政審查程序
仿製藥落入專利範圍	初審行政審查押後至專利到期前才開始
9 個月停留期內未作出裁決	同「落入範圍」處理

The incentive: the first generic to successfully challenge gets 6 months' data exclusivity and 12 months' market exclusivity.

Data exclusivity vs. market exclusivity

Two definitions that are constantly confused Data exclusivity protects the preclinical and clinical trial data used in a regulatory submission, preventing a generic manufacturer from using that data in its own regulatory application. Market exclusivity is a period during which the product is protected from market entry of a generic, hybrid or biosimilar. They are related but distinct: data exclusivity blocks a specific shortcut, market exclusivity blocks entry altogether.

Jurisdiction	Data / market exclusivity periods
China	6 years for pharmaceutical drugs (new chemical entity)
US	5 years for pharmaceutical drugs (new chemical entity) 3 years for new indications of pharmaceutical drugs 12 years for biologics
EU	8 years (+2 years market exclusivity +1 year for new indication)

Compare with L4's numbers and the strategy becomes clear: a biologic in the US has 12 years of exclusivity to protect against, while an NCE has 5 — which is precisely why the biologic's protection strategy leans harder on the patent term and on manufacturing complexity.

Mistake 1 — publishing too early

The hypothetical case (based on a conglomerate of true stories) Dr. Lim, based in Hong Kong, spends 3 months in Shenzhen advising experiments, analysing data and guiding R&D direction. The invention is completed. The timeline then goes: academic publication → private sales $$ → product launch → IPO $$$, while the filings are CN invention patent, US provisional, PCT, and national filings across Europe, Canada, Australia, Korea, Japan and India. The second version of the slide adds the 12-month PCT window and a China invention patent — and crosses out the two countries where nothing was filed first. The problem is not the sequence of events; it is the gap between the publication and the first filing.

The four common mistakes named on the slide Not being careful what you say publicly; publishing or selling too early; statements that make your invention seem “obvious”; inconsistent statements. That list recurs later with a different set (filing strategy, ownership, value), which is itself a useful exam structure: the lecture is organised as four mistake-clusters.

How a publication can hurt a patent

As prior art, making your later-filed application not novel.
As prior teaching, making your later-filed application seem obvious.
If it makes inconsistent statements, it calls into question integrity — i.e. fraud.

Things to consider before disclosing: can I “hide” (redact) the key inventive portions from others until I file? Can I file a patent application first before disclosing the invention? Are my statements inconsistent with any positions we are taking?

The spectrum slide — what counts as “obvious” A gradient runs from original ground-breaking idea → next rational step → improvement → different / unobvious, unexpected, surprising, ground-breaking idea. Where a disclosure sits on this line decides whether it destroys novelty (left end) or merely weakens non-obviousness (middle). The wording you use when you publish places your invention on this line, which is why the next slide is about language.

How to minimise a publication's harm Avoid “progression” language that implies obviousness — words such as “reasonably expect”, “we predict”, “obviously”, “consistent with [what was done before]”. Instead, focus on the unexpected findings. Avoid characterising or taking a position on the prior art — stay factual if possible. Ideally, coordinate the IP strategy together with the publication / public disclosure strategy. The slide's summary is the rule: Less Opinion, More Facts.

Case — Belcher Pharma v. Hospira (2021): inequitable conduct

Allegation: Belcher's US Patent No. 9,283,197 was invalid. The facts turn on a pH window and on the difference between what was said to the FDA and what was said to the USPTO.

	Statement to FDA in an NDA	Statement to USPTO Examiner
Prior art	Belcher filed an NDA on a new epinephrine formulation with a pH range of 2.4 to 2.6, improved over the “old” formulation at pH 2.8–3.3	Prior art cited: Helenek, pH range 2.2 to 5.0
Characterisation	FDA questioned whether changing pH would lead to higher racemization. Belcher cited prior art “Stepensky”: racemization in epinephrine is a “well known process”	Belcher: the pH range 2.8–3.3 was “unexpectedly found to be critical” in reducing racemization of l-epinephrine, and “produced unexpected results”
Why it mattered	The prior art was NOT CITED IN IDS, and the two descriptions are irreconcilable: “well known” and “unexpected” cannot both be true.

The pH 2.8–3.3 range (with oxidation above and racemization below, optimal at 3.0–3.8) was the old marketed product's — Sintetica's — range. Belcher eventually chose to go with that art-recognized pH to get faster FDA approval.

Outcome and the exam point Patent ruled INVALID based on inequitable conduct by the CSO. The lesson is not about chemistry — it is that a statement made in a regulatory filing can destroy a patent in a later prosecution, because the two documents are read together and any inconsistency is read as deliberate. Draft every regulatory and patent document against the others.

Case — Bristol-Myers Squibb v. Rhone-Poulenc Rorer (326 F.3d 1226, 2003): the paper kills the patent

The clearest possible illustration of the timing point, and it is worth memorising because it is a documented, citable instance of the same mistake the hypothetical Case 1 illustrates.

RPR invented an s molecule that would become Taxol.
RPR filed a patent application based on its laboratory manual, the “Modes Operatoires”.
Its JACS paper was published on 17 August 1988; the patent application was filed on 3 April 1989 — roughly seven months later.
The manual recommended using TMS (trimethylsilyl) as a hydroxy-protecting esterification group; the paper said [TMS] proved too fragile in the esterifying process and that the MOM protecting group could not be removed after esterification.
The application said “preferably R3 is [MOM] … preferably, R3 is [TMS] …” The paper said TBDMS could work but TBDM-Cl did not. [TMS] was NOT CITED IN IDS.
PATENT WAS GRANTED — and subsequently invalidated.

Why this case is the exam answer The prior art was not a stranger's paper; it was the inventors' own publication, describing their own preferred embodiment as unworkable. A patent cannot be held by the person whose own paper teaches the public to do what the patent forbids — and the grant does not cure it: the patent was granted and still invalidated, which is exactly the point L6 keeps returning to.

Case — Broad v. University of California: the PTAB reasoning and its evidence

L5 covered the outcome. This lecture gives the reasoning, and the reasoning is the examinable part.

Broad's motion at PTAB: no interference-in-fact, on a preponderance of the evidence, showing that neither side, if considered prior art, would anticipate or render obvious the other.
Is there an interference? Two patentably indistinct inventions? Novelty: YES, NOVEL due to the new feature “eukaryotic cells”. Obviousness: would one of skill in the art have had a reasonable expectation of success?

Three kinds of prior-art evidence

Evidence	Weight / content
The Jinek publication	UC Berkeley 的旗艦論文，顯示 CRISPR-Cas9 在原核細胞中有效
Contemporaneous statements by the inventors	權重高於專家證人——“這是一個大成功，但有個問題：我們不確定 CRISPR/Cas9 在真核細胞（植物與動物細胞）中可行”
Doudna's own words	在讓 CRISPR 在人細胞中運作上經歷「many frustrations」，並說若成功將是 “a profound discovery”；又說在動物與人類中做這些修飾的技術「一直是研究與人類治療開發的巨大瓶頸」

UC's expert witness paper argued that ZFN and TALE were known in the art for eukaryotic systems; Broad pointed to UC's own statement that the components came from natural transcription factors binding in a chromatin context, that Cas9 had no such guarantee, and that “only attempts to apply the system in eukaryotes will address these concerns”.

The distinction that decided the case UC argued ZFN and TALEN were able to “convert” to eukaryotes, and that group II introns worked in eukaryotic cells. Broad's answer is the distinction the PTAB adopted: “motivated to try” is not “reasonable expectation of success.” Chromatin vs. riboswitches is a different problem from ribozymes; needing unique conditions is not the same as being a simpler system. The court kept returning to UC's own expert witness's prior statement.

The PTAB final decision — quotable The preponderance of the evidence, including the contemporaneous statements of the inventors and others in the field, as well as the knowledge of ordinarily skilled artisans, demonstrates that one of ordinary skill would not have had a reasonable expectation of success that CRISPR-Cas9 could be used in a eukaryotic cell. If they were prior art, UC's claims would not have rendered Broad's obvious; nor would they anticipate Broad's. Therefore the claims are not drawn to the same patentable subject matter and there is no interference-in-fact. The PTAB also noted that “a narrow species can be non-obvious and patent eligible despite a patent on its genus” — which links straight back to the genus/species drafting discussion in L2 and L5. UC appealed to the Federal Court but was denied.

Mistake 2 — no filing strategy

The second mistake cluster: lack of a patent filing strategy; filing IP at the wrong time (too early / too late); filing in the wrong jurisdiction first; and becoming one's own prior art. That last phrase is the whole Belcher and BMS story in four words, and it is the most quotable line in the lecture.

Export control and security clearance: where the invention was made actually bites

	CHINA	MACAU	HONG KONG	US
Legal family	Civil law	Civil law	Common law	Common law
Derived from	German law	Portuguese law	British law	US law
Export controls on inventions	有	無此類管制	無此類管制	有
Framing	“Completed in China”	—	—	“Made in the US”

Why this matters more than it looks Because Hong Kong and Macau have no such controls, a Hong Kong R&D site is often treated as “outside China” in casual advice — but an invention substantively completed in Shenzhen is an invention completed in China, wherever the team sat and however the collaboration was structured. The hypothetical Case 1 is built exactly on this: Hong Kong HQ, three months in Shenzhen, invention completed.

Collaborating with China: arising IP

Inventions “completed in China” must be filed in China FIRST, or a foreign filing licence must be obtained.
Does the CRO own the IP, or do you?
Do the inventors' former employers have any rights to the arising IP as service inventions?

Foreign filing licence (FFL)

Requirement	Detail
Filing requirements	“Completed in China” 的發明若先往國外申請，必須先通過保密審查；未通過者在中國無法取得授權專利
Option 1	以中文提交說明書供審查；實務週期相對快（1–2 週）
Option 2	以中國為受理局提交 PCT —— 自動視為提出保密審查請求；可用英文（不同於直接的中國申請）；須符合在 CN／RO 申請的資格
Hard limit	不能事後補辦 foreign filing licence
Practical	外國人須透過中國專利代理機構辦理

Jiecang — patent invalidated for failure to obtain an FFL 2022 case, CN Patent 201720389490.8 (the “'490 patent”). The timeline:

Date	Event
Before	發明人Renchang Hu travelled to the US 10 days before the US Provisional application was filed
Dec 20, 2016	US Provisional application
Jan 10, 2017	CN Priority Filing
Apr 14, 2017	“the '490 patent”

Other facts: all production facilities were in China; there was no evidence of the 3 other listed co-inventors leaving China. Invalidated on the reasoning that “there was a high probability that the substantive content of the invention was completed in China”, and it was not possible to overcome the technical barriers with a production facility — i.e. the person could have built it in China without travelling. The rule is about where the work happened, not where the inventor happened to be on a given day.

Mistake 3 — ownership and inventorship

The third cluster: ownership of invention; not fully considering what affects patent ownership; not considering the consequences of jointly developed inventions; and the need for good agreements in place. The slide's framing question is blunt: who owns the arising IP? The determinants are listed as: where the invention was made; who made it; who they worked for at the time; which country's laws apply; which contracts were in place; and which contract trumps which.

Inventorship is not authorship

Two different legal determinations Who is an inventor? — a legal determination: one who conceives of the invention. Conception is the formation of a definite and permanent idea of a complete and operable invention such as is claimed in a patent. Who is the owner? — in the US, ownership follows inventorship (so be careful about co-inventing with others); in China it may follow employment. The slide's punchline: inventorship is not the same as authorship — being the person who wrote it, or the senior person who directed it, or the lab head, does not make you an inventor; and being an inventor does not automatically make you the owner.

Why getting it wrong is fatal The slide gives three consequences directly: wrong inventors → possibly an invalid patent; ownership errors carry legal consequences; and inventorship affects what counts as prior art — if your own co-inventor is named wrongly, the prior-art analysis and the chain of title both fail at once.

SiOnyx v. Hamamatsu Photonics — contract trumps patent law Two parallel tracks. SiOnyx (US company) filed a US provisional on 25 May 2001 and disclosed SiOnyx confidential information to Hamamatsu (JP company) at a Nov 2006 meeting under an NDA stating: “[t]he Receiving Party acknowledges that the Disclosing Party … claims ownership of the Confidential Information disclosed by the Disclosing Party and all patent, copyright, trademark, trade secret, and other intellectual property rights in, or arising from, such Confidential Information.” The NDA expired Jan 2008. In Feb 2009 Hamamatsu filed a JP patent application; it was the main inventor but referred to prototypes from before. Hamamatsu co-invented with its own employees and others, then pursued US / PCT / EP filings. Verdict: SiOnyx owns ALL the patents, even the foreign ones — because the evidence showed the relevant information came solely from SiOnyx's confidential information.

The sentence to remember PATENT LAW: co-invention should be JOINTLY OWNED — BUT a contract can TRUMP patent law. This is the joint-inventorship counterexample that every cross-border biotech needs: had the NDA not contained an express assignment of “all IP arising from” the disclosed information, the joint-inventorship rule would have put the Japanese company on the patent.

Mistake 4 — the value of IP, and the due-diligence hole

The fourth cluster: value of IP; not willing to invest properly in IP until it's too late; should have a separate budget for it, like R&D; and due diligence and freedom to operate.

The hypothetical Case 1 returns with the fixed version: a 12-month PCT window marked, and a China invention patent filed first, before the US provisional and the PCT. The version with the X is the failure case; this one is the corrected sequence.

Why the ordering is CN first Because an invention substantively completed in China must be filed in China first or require an FFL — so “file in China first” is not a China-centric preference, it is a validity condition. A US-first strategy, however convenient for the US market, risks the China patent being invalid (as in Jiecang) even when everything else is done correctly.

Key takeaways

Don't publicly disclose until you file — and disclose with care when you do.
Invest in a smart filing strategy early on.
Remember that “filing a patent” (an application) is not the same as getting a granted patent with strong, broad and enforceable claims.
Healthcare / biopharma IP has its own set of unique challenges that require a specialist.

One-line Summary English: Strategic IP is a set of coordinated business decisions — what your objective is, where the invention was substantively made, where the commercial footprint is, and when the public events fall — and this lecture learns the discipline through four mistake-clusters: publishing before filing (Belcher's pH window contradicted across an NDA and a prosecution; BMS's Taxol application destroyed by its own JACS paper), filing without strategy in the wrong jurisdiction first (Jiecang invalidated because the invention was completed in China), getting ownership wrong (SiOnyx won the foreign patents because the NDA assigned “all IP arising from” the disclosed information), and underinvesting in IP until due diligence exposes the gap.
中文邏輯：策略性 IP 是一組互相牽動的商業決策——目標是什麼、發明實質上在哪裡做成、商業足跡在哪裡、以及公開事件落在什麼時點——而本講用四組錯誤學會這套紀律：先公開後申請（Belcher 的 pH 範圍在NDA 與專利審查中說法互相矛盾；BMS 的 Taxol 申請被自己的 JACS 論文摧毀）、沒有策略而在錯誤法域先申請（Jiecang 因發明在中國完成而被無效）、產權搞錯（SiOnyx 靠 NDA 中「所有由此產生的 IP」條款拿下外國專利）、以及在盡調才發現缺口時才願意投入 IP。