Keymed’s CM512 meets its Phase 2 endpoints: an announcement without numbers, and a selling point already taken

The prevalence of chronic rhinosinusitis in China is often cited at about 107 million people, about a third of whom have nasal polyps, and around 80% of whom are type 2 inflammation-driven. The typical course for these patients: limited effect from nasal steroids, oral steroids only for short-term use, a high rate of recurrence even after surgery, and repeat surgery not uncommon. Biologics are intended precisely for this difficult-to-treat population.

But stretch the timeline to August 2026, and the description of “unmet need” needs updating. China’s CRSwNP field now has five approved biologics:

The last row is key. GSK’s depemokimab was approved in China for CRSwNP in April 2026 with a 26-week dosing interval — twice a year. In other words, the “one injection every six months” position had already been taken before CM512 read out its Phase 2 results.

This has a big impact on CM512’s value positioning. If long action is its only differentiation, then on entering the market it faces not a group of every-two-weeks products, but an incumbent that is also twice a year, launched three to four years earlier, and very likely already has NRDL access. Differentiation must land somewhere else — either magnitude of efficacy or breadth of the responding population.

And depemokimab happens to leave an opening on efficacy magnitude. In its two Phase 3 studies, ANCHOR-1 and ANCHOR-2, change from baseline in nasal polyp score at 52 weeks was −0.7 and −0.6 respectively, and the nasal congestion score (VRS) at Weeks 49–52 was −0.23 and −0.25. This magnitude is relatively modest, clearly below the level of IL-4Rα products. In theory, a bispecific suppressing upstream and downstream simultaneously has a chance to achieve efficacy on the scale of IL-4Rα products while keeping twice-yearly dosing — that would be real differentiation.

That is exactly the problem.

The announcement gave key information at the design level: three treatment arms of 300 mg Q12W, 300 mg Q24W and 600 mg Q24W, plus a placebo control; a 24-week treatment period followed by 12 weeks of safety follow-up; and the primary endpoint described as change from baseline in nasal polyp score at 24 weeks being “significantly superior to the control group”.

But the announcement did not disclose any specific values for nasal polyp score or nasal congestion score — no between-group differences, no confidence intervals, no p-values, and not even the number of patients enrolled.

Combined with registry information, this study (NEZHA-1) is a randomized, quadruple-blind, placebo-controlled parallel design with 120 planned patients; eligibility requires a nasal polyp score of at least 5 with at least 2 on each side and a nasal congestion score of 2 to 3, and the registered primary endpoint is change from baseline in nasal polyp score at Week 24.

Putting these together, several things can be seen:

First, “met all study endpoints” and “an effect size large enough to beat the incumbents” are two different propositions. The former only shows superiority to placebo; the latter needs specific numbers. Until the between-group difference is available, there is no way to tell whether CM512 is close to the level of IL-4Rα products or falls in depemokimab’s more modest range. The asset values corresponding to these two scenarios are very far apart.

Second, the statement “comparable efficacy across dose arms” carries more weight than usual under this dose-arm design. Because the three treatment arms include both 300 mg every 12 weeks and 300 mg every 24 weeks — the same dose with intervals differing twofold. Comparable efficacy at 24 weeks means halving the dosing frequency came at no visible efficacy cost; this is the most direct support for “once every six months”, not just PK extrapolation. Of course there is another explanation: with about 30 patients per arm, the sample may simply be too small to distinguish differences between arms, in which case neither the minimum effective dose nor the optimal interval has really been defined yet.

Third, follow-up length is still a weakness, but not as short as one might think. Week 24 is exactly the end of the first complete dosing cycle for the 300 mg Q24W arm, meaning this endpoint was measured “before the next injection”, so efficacy at trough has been tested once. What is really missing is data spanning two or more dosing cycles — registration studies of the incumbents generally already have 52-week efficacy and safety results, and depemokimab’s Phase 3 primary endpoint was itself set at 52 weeks. The open-label extension study NEZHA-2 will follow patients to Week 60, but results won’t be available until 2028.

For comparison, CM512’s published Phase 1 data are relatively solid: of 110 subjects, 46 were patients with moderate-to-severe atopic dermatitis; in the 300 mg arm, the Week 12 EASI-75 response rate was 58.3% and EASI-90 41.7%, versus 21.4% and 0% on placebo; at Week 6, EASI-75 was 50% vs 7%. On safety, there were no serious adverse events and no discontinuations due to adverse events, treatment-related adverse events were all grade 1–2, and anti-drug antibodies were 3.1% in the single-dose arms and 0% in the multiple-dose arms. Note that hypertriglyceridemia occurred in 28.1% of the single-dose arms versus 12.5% on placebo, which merits long-term follow-up.

But the Phase 1 data come from atopic dermatitis, not nasal polyps; and the denominators are small — about 12 patients in the 300 mg arm and about 14 on placebo, so point estimates are not stable. Cross-indication, small-sample data cannot replace effect sizes in the target indication.

CM512 was added to the Breakthrough Therapy list on July 23, 2026 — a genuine positive, meaning enhanced communication with regulators and eligibility to apply for priority review when filing the marketing application. But one must be clear-eyed: Breakthrough Therapy and conditional approval are two different procedures; the latter requires statutory conditions such as a seriously life-threatening disease with no effective treatment, and nasal polyps do not qualify, with five biologics already approved in China — against this background, the realistic likelihood of conditional approval is low. The value of acceleration most likely lies in the review clock rather than in reduced evidence requirements, and a Phase 3 can hardly be waived.

And on the road to Phase 3, the squeeze comes from three directions.

Stacking these up, a reasonable timeline projection: Phase 3 first patient dosed in Q4 2026 to Q1 2027, 24-week primary endpoint topline around 2028, 52-week data around 2029, marketing application in 2028–2029, and an approval window roughly in 2030–2031. By then CM512 will be the sixth or later biologic in China’s CRSwNP field.

To discuss CM512’s transaction value, one thing must be clarified first: its rights are no longer complete.

What needs to be understood precisely is the scope of this license: Belenos was granted exclusive rights to research, develop, register, manufacture and commercialize outside Greater China — that is, Belenos itself holds manufacturing rights within its licensed territory. The separate supply arrangement between the two gave Belenos an option: it may purchase CM512 and CM536 needed for clinical trials from Chengdu Keymed or its contract manufacturers. This is not the same as “the seller retains manufacturing”, and one cannot infer from it that Keymed never does technology transfer in external collaborations.

This means several things. First, any collaboration on CM512 can only address Greater China. Second, Belenos is already advancing CM512 (its code BEL512) in the clinic in the US, with asthma rather than nasal polyps as its priority indication — the China and US development paths are in fact proceeding separately; all sites for the five Phase 2 studies in China are domestic, with no overseas centers and no multi-regional trials. Third, there is a technical but not minor framing issue: the HKEX announcement uses “Greater China region”, while Belenos’s own press release says “all ex-China rights”. The two do not match literally, and ownership of Hong Kong, Macau and Taiwan needs to be determined by the definitions clause in the agreement text.

Another issue not yet public but affecting real economics: are the heterodimerization technology and Fc half-life extension used in CM512 proprietary or licensed from a third party? Neither of the two published peer-reviewed papers discloses this. Keymed’s publicly traceable in-house bispecific platform is nTCE, aimed at T-cell engagers, which is not the same format as a cytokine-neutralizing 1+1 IgG like CM512, so it cannot be presumed to be proprietary technology. If third-party patents are involved, they would create an additional layer of royalty burden.

Public information on manufacturing is quite limited. In the first public notice of the environmental impact assessment for Chengdu Kangnuoxing’s innovative antibody industrialization project on June 11, 2026, the existing engineering products listed included CM310, CM338, CM326, CM313, CM336, CM355 and stapokibart, but not CM512; the new production line is still at the EIA notice stage. It should be noted that not being listed as an existing engineering product in one EIA document only shows that document did not list it, and is not enough to infer the actual progress of commercial process scale-up and registration batch preparation. Current public materials are insufficient to judge CM512’s commercial capacity arrangements and readiness of the registration batch process.

To find pricing coordinates for CM512’s Greater China rights, the most persuasive references are deals on a consistent basis — that is, autoimmune biologics deals likewise transferring China or Greater China rights to domestic companies. There are three such deals for reference:

All three deals transferred China / Greater China rights to domestic companies. Upfronts are concentrated in the US$10–21 million range.

First, Keymed’s own CM326, licensed to CSPC in November 2021, with an upfront of RMB 100 million and development milestones up to RMB 100 million, plus sales milestones and royalties with amounts/rates not disclosed. This is a direct observation of the same seller giving out China rights.

Second, Simcere’s in-licensing of Greater China rights to Connect Biopharma’s rademikibart in November 2023, when the asset was at the pre-marketing-application stage: an upfront of RMB 150 million (about US$21 million), milestones up to about US$120 million, and royalties up to the low double digits in percent.

Third, Hansoh’s in-licensing of Greater China rights to Qyuns Therapeutics’ QX004N in April 2024, at Phase 2 at signing: an upfront of RMB 75 million (about US$10.4 million), milestones up to about US$142.5 million, and tiered royalties from high single digits to double digits.

The upfronts of the three deals are concentrated in the US$10–21 million range. With Phase 2 meeting all endpoints, Breakthrough Therapy Designation and Phase 3 imminent, CM512 should be able to aim for the upper end of this range or even a modest premium; but given that the effect size has not been disclosed and the long-acting selling point has been taken by an incumbent, room for a premium is limited, and an order-of-magnitude jump is unlikely.

In contrast is a deal in the other direction. Hengrui’s long-acting TSLP antibody SHR-1905 had its ex-Greater China rights licensed out in 2023 for a US$25 million upfront; the licensee, Aiolos, was then acquired by GSK in January 2024 for US$1 billion upfront plus US$400 million in regulatory milestones. This figure is often cited, but when citing it note: what GSK bought likewise excluded Greater China; the China rights remain with Hengrui. Overseas rights and China rights are assets on two different bases, and their prices are not interchangeable. CM512 is almost a structural mirror image — overseas has already gone to Belenos, and what stays at home is the China half.

As for strategic recognition of the mechanism, in June 2026 AbbVie announced the acquisition of Apogee Therapeutics at an equity value of about US$10.9 billion (enterprise value not disclosed in the announcement; the target’s net cash would need to be deducted separately). APG273 in its pipeline is precisely a combination of long-acting IL-13 and long-acting TSLP, intended for dosing every quarter or every six months — functionally very close to CM512, the difference being that AbbVie’s route combines two monoclonal antibodies while CM512 is a single-molecule bispecific. This shows big multinationals recognize this mechanistic direction, but the price scale of that deal is not comparable with a China-rights asset.

There is also an often-overlooked dimension: whether Keymed has any motive to part with its China rights.

Financially, there is almost no pressure. At the end of 2025, the company’s cash and bank balances, time deposits, restricted cash and wealth management products totaled about RMB 1.963 billion; on June 5, 2026, the acquisition by Gilead of Ouro Medicines, in which it held a minority stake, closed, and the company announced receipt of US$257 million; in March of the same year it also received the group’s share of a US$45 million milestone from AstraZeneca triggered by first patient dosed in the CMG901 Phase 3. A positive profit alert on July 15, 2026 indicated first-half profit of no less than RMB 1.2 billion.

Strategically, the motive may even run the other way. CM512 covers five type 2 inflammatory indications — nasal polyps, atopic dermatitis, asthma, COPD and chronic spontaneous urticaria — while Keymed already has a commercial team of more than 400 people selling stapokibart, which likewise covers atopic dermatitis, nasal polyps and seasonal allergic rhinitis, with all three indications on the NRDL. CM512 is the most natural next-generation successor for this product line.

Then look at the pattern of behavior. Since February 2023, Keymed has completed five out-licensing deals involving six assets — CMG901 to AstraZeneca, CM512 and CM536 bundled to Belenos (in the same agreement), CM336 to Ouro, CM313 to Timberlyne, and CM355 to Prolium — with a highly consistent feature: a low cash upfront, paired with a large stake in a new company, milestones and tiered royalties, R&D support service fees and supply agreements, and always retaining Greater China. The only one that gave away China rights was CM326 to CSPC.

In other words, making an offer for CM512’s China rights runs counter to the consistent preference shown in this company’s five recent out-licensing deals. That doesn’t mean it is impossible, but it does mean the first step in any discussion is not valuation, but confirming whether the other side really wants to talk.

Summary. CM512 is an asset with good completeness at the molecular level: both targets independently validated, a pharmacodynamic chain already closed in humans, and a 70-day half-life supporting the real convenience of two injections a year. The Phase 2 dose-arm design included both 300 mg every 12 weeks and every 24 weeks, with comparable efficacy, turning “once every six months” from a PK inference into a conclusion backed by clinical comparison.

But it now sits in an awkward position — the easiest story to tell (once every six months) has already been taken by an earlier-approved product, the evidence that best proves its value (effect size on nasal polyp score) has not been published, overseas rights have been licensed out, readiness of commercial capacity cannot be seen clearly at the public level, and the seller is not short of money and has every reason to develop it itself.

Judgment on this asset therefore depends heavily on a number that has not yet appeared. Until that number comes out, any valuation can only be conditional.