Asandeutertinib approved: it won the head-to-head, but arrived two years late to its position

Zhejiang TYK Medicines was founded in November 2017, registered in Changxing, Huzhou, Zhejiang, and listed on the Main Board of the Hong Kong Stock Exchange on August 20, 2024, when the market dubbed it “the first deuterated-drug stock in Hong Kong”. Chairman and CEO Dr. Wu Yusheng was formerly a senior researcher at multinational pharma companies and is a national distinguished expert. As of June 30, 2026, the company had 184 employees (163 a year earlier) and a pipeline of 10 candidates, all small-molecule targeted oncology drugs.

TY-9591’s official Chinese generic name corresponds to asandeutertinib mesylate tablets, English INN asandeutertinib (company announcements during the filing stage used “aiduotinib mesylate tablets”). Its chemical logic is straightforward: replace hydrogen with deuterium at specific positions on the osimertinib molecule. Carbon–deuterium bonds are more stable than carbon–hydrogen bonds, so metabolism is slower. What the company aims to suppress is an active metabolite of osimertinib, AZ5104 — which has higher affinity for normal, non-mutated EGFR and is thought to be the main source of adverse reactions such as rash and diarrhea. Single 80 mg dose data in healthy volunteers showed exposure to this metabolite about 50% lower than with osimertinib.

Fewer metabolites should in theory mean better tolerability, allowing the dose to be pushed up. Osimertinib is 80 mg a day; asandeutertinib is 160 mg a day. That is the company’s design intent; however, it should be noted that human brain exposure and cerebrospinal fluid concentration data have not been published, so the causal chain “higher dose → higher brain exposure → intracranial superiority” remains an inference; what has actually been proven is only the difference on clinical endpoints.

Osimertinib’s compound patent in China expires on July 25, 2032. The first domestic generic was approved back in October 2023, but because the patent is still in force, it effectively cannot be sold. This means that until 2032, the price-setters in this race remain the originator plus NRDL-listed domestic third-generation drugs, not generics.

China is the world’s most concentrated market for EGFR-mutant lung cancer. According to Frost & Sullivan, China had about 236,000 newly diagnosed advanced and early-recurrent EGFR-sensitizing-mutation non-small cell lung cancer patients in 2023, projected to grow to 270,000 by 2030; EGFR mutations account for about 51% of non-small cell lung cancer in China, far higher than in Western populations. These patients are predominantly non-smokers, adenocarcinoma and female, overlapping heavily with the ESAONA enrolled population.

Brain metastasis is an unavoidable problem for this group. The incidence of central nervous system metastases in EGFR-mutant patients is significantly higher than in other driver-gene populations, and intracranial progression often precedes systemic progression, a key event determining quality of life and survival. Past evidence for third-generation drugs in brain metastasis populations came almost entirely from subgroup analyses of main studies or small exploratory studies, of low evidence level.

So “a registration study designed specifically for brain metastases, with an intracranial endpoint as primary endpoint” was indeed a real gap. The problem is — this gap was already filled once in November 2024.

There is a third party in the same position too. In June 2024, osimertinib combined with pemetrexed and platinum was approved in China for first line, with more pronounced benefit in patients with central nervous system metastases and L858R mutations. In other words, for a patient with brain metastases, clinicians today already have three labeled paths: osimertinib monotherapy, osimertinib plus chemotherapy, and zorifertinib. Asandeutertinib is trying to be the fourth.

More broadly, China has approved 7 third-generation EGFR-TKIs, with more than 10 others in Phase 3. This is, bar none, the most fiercely competitive targeted therapy race in China.

Between “good data” and “it sells”, this asset faces seven hard constraints. If any one collapses, the projections that follow don’t hold.

Laying out all of TY-9591’s clinical activity, one feature is very clear: all study sites are in China — not a single overseas IND, not a single multi-regional study, and no overseas license of any kind. This is not an oversight but a choice locked in by real conditions.

First, why it works in China. First, the population base is here — EGFR mutations account for about 51%, more than twice the Western level, and the brain metastasis population is large enough to support a registration study for a standalone indication; the same study would enroll far more slowly and expensively in Europe and the US. Second, someone has already paved the review path — zorifertinib was approved for an indication with exactly the same wording, showing regulators accept the logic of “intracranial endpoints supporting a brain-metastasis-specific indication”. Third, the expedited channel was obtained — a pre-NDA meeting request was submitted in November 2025, agreement to file came in December, the product was formally added to the priority review list in January 2026, the application was accepted in February and conditionally approved on August 27: about three months from meeting to acceptance and about ten to approval. Fourth, the network of leading investigators is highly concentrated, allowing a 224-patient head-to-head study to complete enrollment and read out within just over two years.

Now, why going abroad doesn’t work. ESAONA is open-label, purely Chinese and uses surrogate primary endpoints. Any one of these three would need extra explanation in an overseas review context. It should be noted that US regulators do not categorically reject purely foreign clinical data — they can be accepted if the data are applicable to the US population and study quality and verifiability requirements are met — so the more accurate judgment is: additional bridging or multi-regional studies are expected to be needed, with specific requirements depending on regulatory discussions. And at the company’s cash scale, an international multicenter Phase 3 would cost the equivalent of spending the past five years’ R&D investment all over again.

The logic of this deal is beyond reproach: a 140-person company with no commercialization experience could not build a nationwide oncology sales network to fight in a market with 7 approved competitors. Bringing in channels is rational.

But the cost must be seen clearly too: a large slice of the core product’s long-term China margin has been cut away, the company’s control over end-market pricing and access pace has declined accordingly, and the money the deal brings in is, as agreed, mainly directed to another product. In the early launch phase of the core product, the company’s own room for marketing investment has actually been compressed. Add the unresolved agency litigation, and the exclusivity of this deal still carries an undefused question mark.

The clinical value of the brain metastasis population is real: intracranial progression often occurs before systemic progression, and once it does, patients’ functional status, quality of life and subsequent treatment options deteriorate simultaneously. Historically these patients were excluded from most registration studies; only in recent years have studies designed specifically for them appeared.

But commercially, this territory has three structural limits that need to be put on the table.

First, a narrow label means a narrow market. The indication is restricted to “with central nervous system metastases”, not the entire EGFR-mutant first-line population. A narrow label can buy faster approval, but it also locks in a smaller population base for NRDL negotiation. The company is clearly aware of this, and is running a separate double-blind Phase 3 in the first-line L858R population — 548 patients enrolled as of April 2026, with full enrollment now expected to slip to Q4 2026, and filing still planned for 2028. The filing date for this study was moved in company disclosures from 2026 to 2028, and the enrollment completion date has also slipped from the original Q2 2026; the broad label will arrive about two years late.

Second, resistance to prescription switching is bigger than the data gap. For a department already using osimertinib, aumolertinib or furmonertinib, switching requires clearing three hurdles: whether the new drug is on the NRDL, patients’ ability to pay, and physicians’ familiarity with the safety of a new molecule. ESAONA’s intracranial data are enough to persuade at the academic level, but a grade ≥3 adverse event rate of nearly 50% will keep a considerable share of physicians cautious in primary-care settings.

Third, the lifecycle is on the short side. Osimertinib has already extended its territory to adjuvant therapy and unresectable stage III, while TY-9591 currently has no perioperative program. The only visible expansion paths are three: the full L858R population, combination with platinum-based chemotherapy (Phase 2 enrollment complete, with preliminary data cleaning and summary completed in Q4 2025), and leptomeningeal metastases. Each requires independent registration investment, and each already has approved competitors standing ahead.

There is also a long-term issue: deuterated improvement does not change EGFR resistance pathways. With longer first-line use, patients after resistance still face the same C797S, MET amplification and other issues, and the company has no corresponding later-line program in public information. This patient journey currently covers only the very first segment.

Manufacturing deuterated small molecules has its peculiarities: supply of deuterated intermediates is highly concentrated, isotopic purity control is key to process scale-up, and raw material costs are clearly higher than for ordinary APIs. Fortunately, raw material costs account for a limited share of oral solid dosage forms, so the overall margin structure remains friendly.

The company’s new solid dosage plant, once built, will have annual capacity of 150 million tablets or capsules; phase one civil works passed completion acceptance in June 2024, but per the latest disclosure, the phase one line will only complete compliance certification and be ready to produce in Q1 2027. And the drug manufacturing license obtained in January 2026 lists Asymchem Life Science (Tianjin) Co., Ltd. as asandeutertinib’s contract manufacturer.
Taken together, these two facts mean: commercial supply in the first year after launch goes through contract manufacturing, and own production lines are a second phase that only picks up after 2027. For a company that has just received its first approval, this arrangement is pragmatic, but it also means capacity, cost and release pace in the early launch phase are in the hands of an external partner.

With the Qilu deal layered on, the supply structure becomes three-part: API development and manufacturing rights granted to Qilu, formulation manufactured by Asymchem Tianjin under contract in the early launch phase, and marketing authorization holder status and the selling entity kept by the company itself. The benefit is being asset-light and fast to market; the risk is two external single points in the chain, and a problem at either end — including legal disputes affecting the stability of cooperation — would directly hit supply continuity. Establishing a second-source contingency plan and clear quality agreements should be a must-do within the next year.

Overlay the three timelines of money, review and access, and the problem becomes immediately clear:

The most dangerous stretch in this table is the dozen-plus months between approval in August 2026 and NRDL implementation in January 2028 at the earliest. The product is on the market, but has to compete at a self-pay price against NRDL-listed third-generation drugs costing a few thousand yuan a month. In company NRDL submissions, osimertinib is described as having the lowest average monthly treatment cost among same-class third-generation products, a very low price anchor. If asandeutertinib is priced close to competitors, margins will be squeezed from both sides by promotion fees and deuterated raw material costs; if priced significantly higher, penetration will be extremely limited.

This also explains the practical meaning of the RMB 2.06 billion in milestones in the Qilu deal: it is a long-term cap, not near-term cash flow. What might actually be banked in the next two years is the roughly RMB 700 million in upfront consideration (of which the RMB 400 million subscription is still to close) plus the first few milestone tiers linked to approval and indication expansion. Until the subscription is completed, this money can only be modeled as “agreed”, not “received”.

TY-9591 is a very typical Chinese-style improved asset: not disruptive scientifically, but very solidly engineered; using the hardest design — head-to-head against the current standard of care — it obtained data good enough for an oral presentation at an international congress. These data themselves deserve credit; they prove a domestic company can win once in the most crowded race using the most rigorous method.

But it is also a reminder that in today’s Chinese oncology market, more and more stands between “better data” and “better sales”: a predecessor approved two years earlier and already in guidelines, a grade ≥3 event rate of nearly 50%, a 12- to 18-month self-pay window, a lawsuit over agency rights, and a balance sheet that had to sell its China rights to complete the last mile.

The approval is in hand, so the questions at the table have changed. The three things most worth watching over the next 12 months are: whether the RMB 400 million subscription closes before September 20, how the agency rights litigation is resolved, and first-year uptake and commercial insurance coverage during the self-pay period; a year after that, the readout of the full L858R population Phase 3. The first determines how much cash it has, the second whether its sales channel is clean, and the third how big the table really is.