CNTB / rademikibart-acute-asthma-exacerbation — Rademikibart for acute exacerbations of asthma with type 2 inflammation

Program analysis · bpiq_drug_id 13433 · prepared 2026-08 · USD · framework v5.6.1 · Coverage: CLEARED

Company context, financials, ownership, options and the full pipeline: ../company.md.

How to read this. Written for a reader who has not studied pharmacology or finance. Terms specific to this program are in the Glossary below; general terms are defined once in framework/04-glossary.md. Tags: [VERIFIED — source] / [UNVERIFIED] / [WEB ESTIMATE — source, date].

Glossary

Executive summary

• What it is (one sentence): A single subcutaneous injection of rademikibart, an antibody that blocks the IL-4Rα receptor roughly twice as tightly as the marketed drug dupilumab, given on top of standard steroid treatment to patients who have just turned up at urgent care with an asthma attack and who have high blood eosinophil counts.
• The event and when (as disclosed): Topline results from the Phase 2 Seabreeze STAT Asthma trial (NCT06940141, n=160). The company’s own most recent wording is “September 2026”, stated 2026-08-12; the third-party feed carries “Early September 2026”. No source names a day. This analysis’s readout window is 2026-09-01 to 2026-10-31, likeliest 2026-09-15, at MONTH precision and MEDIUM confidence.
• The main reason it could work: Nothing is approved for this setting, and rademikibart is unusually fast. In its own earlier Phase 2b asthma trial it lifted lung function measurably within the first week and by 420 mL above placebo at Week 24 in exactly the high-eosinophil patients this trial selects for; in a Phase 1 intravenous study it produced 200–400 mL gains starting within 15 minutes. Speed is the property the acute setting needs, and it is the property the class’s existing drug was never developed to have.
• The main risk: Every patient in the trial also gets systemic steroids, which suppress the same inflammation rademikibart blocks, and with 80 patients per arm the trial can only prove a large effect on its 28-day treatment-failure endpoint. The company’s own framing — that the topline “should help determine a potential Phase 3 endpoint and sample size requirements” — reads like a study designed to inform a Phase 3 rather than one expected to deliver a clean statistical win.
• What it means for the stock: This is the dominant program on a $115M enterprise value, so the result reprices the company — but the COPD half of the same trial programme reads out in the same month, so no price move in the window is attributable to this result alone. The scored calls below are miss on the outcome at 38% and up on the shares at low confidence, and those two are not contradictory: the payoff is asymmetric and the market will read the lung-function secondary and the Phase 3 path, not just the strict 28-day composite.

0. Program-tier coverage — CLEARED

One row per mandatory tool in the program-tier table in 02-connectors.md, in that file’s order. Company-tier coverage is in ../company.md C.0. Two mandatory rows are BLOCKED after the full retry policy; none is NOT CALLED, so the gate in 02-connectors.md § The gate is passed and this program is CLEARED. What each block costs is stated in its own row rather than left for the reader to work out.

The regulatory tier was not called at all, and that is correct rather than a gap: the tier is conditional on a regulatory catalyst, and this catalyst is a trial readout.

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

What the drug is. Rademikibart is a fully human monoclonal antibody — a laboratory-made protein built to the same design as a natural human antibody — that binds to a receptor subunit called interleukin-4 receptor alpha (IL-4Rα) on the surface of cells [VERIFIED — Scientific Reports 2023, DOI 10.1038/s41598-023-39311-2, via PubMed 37524768]. In this trial it is given as a single 600 mg injection under the skin [VERIFIED — exhibit 99.1 to CNTB Form 8-K, 2026-08-12; CT.gov NCT06940141 lists "Rademikibart in prefilled syringe" against a matching placebo].

How it works. During an allergic asthma attack, immune cells release two signalling proteins, interleukin-4 and interleukin-13. Neither can act on a cell without first docking at IL-4Rα, which both of them share. Blocking that one subunit therefore switches off both proteins at once, and with them the mucus production, airway swelling and inflammatory-cell recruitment that narrow the airway. Internally, the blocked signal is the one that would otherwise have travelled through a messenger called STAT6 to the cell’s genes [VERIFIED — Scientific Reports 2023, DOI 10.1038/s41598-023-39311-2].

How well the target is validated. Very well, and by the strongest possible kind of evidence: another drug hitting the same receptor is already approved. Dupilumab, also an IL-4Rα antibody, has been approved for maintenance treatment of eosinophilic asthma since 2018 and for chronic obstructive pulmonary disease since 2024, across eight indications in total [WEB ESTIMATE — dupilumab competitive-landscape review via web search, 2026-08-13]. Rademikibart has also produced its own randomised, placebo-controlled, published evidence in asthma — see A.2.

Note what is missing from that validation, because the gap is real: independent human-genetics support for IL-4Rα could not be obtained this session. The Open Targets connector returned Rate limit exceeded for client: global on all three attempts (section 0), so the genetic side of target validation is [UNVERIFIED]. Given that an approved drug already blocks this receptor in this disease, the genetics would be confirmatory rather than decisive.

Where rademikibart claims to differ from dupilumab. Three published, quantified differences, all in the same direction:

1. Tighter binding. Measured dissociation constants of 20.7 picomolar for rademikibart against 45.8 picomolar for dupilumab — a lower number means tighter binding, so rademikibart binds about 2.2× more tightly, at a different epitope on the same receptor [VERIFIED — Scientific Reports 2023, DOI 10.1038/s41598-023-39311-2].
2. More complete signal shutdown. Rademikibart was significantly more potent than dupilumab at blocking STAT6 signalling driven by IL-4 (p < 0.01) and by IL-13 (p = 0.03), with non-significant trends in the same direction on two other assays and equal potency on a third [VERIFIED — same paper].
3. A structural explanation for both. A 2.71 Å crystal structure of the rademikibart antibody fragment bound to IL-4Rα, compared against the 2.82 Å dupilumab structure, shows the two antibodies bound 54.88° apart. That rotation puts rademikibart’s contact surface closer to the interface IL-4 and IL-13 themselves use, and adds a direct interaction with a receptor loop (residues 145–153) that dupilumab does not touch [UNVERIFIED — bioRxiv preprint 2026-04-13, DOI 10.64898/2026.04.12.718052; a preprint has not been peer-reviewed, which is why this is tagged UNVERIFIED while the two rows above are not].

One further difference matters for safety rather than potency. IL-4Rα blockade is known to raise blood eosinophil counts as a class effect, which was seen in dupilumab’s Phase 3 programme. In rademikibart’s own Phase 2b asthma trial, mean eosinophil counts fell rather than rose — by 25 cells/µL at Week 12 and 97 cells/µL at Week 24 in the 300 mg group, against a placebo change of +2 and −23 — and among patients starting above 500 cells/µL, fewer rademikibart patients than placebo patients exceeded 1,500 cells/µL (10.0% versus 18.8%) [VERIFIED — Chest 2026, DOI 10.1016/j.chest.2026.04.019, via PubMed 42061700]. The authors’ own conclusion is appropriately hedged: rademikibart “may not result in hypereosinophilia; however, larger confirmatory analyses are required.”

The exact scientific step this readout must prove. Not that IL-4Rα blockade works in asthma — that is settled. Not that rademikibart blocks IL-4Rα — that is settled too. The step is narrower and harder: that blocking type 2 inflammation adds measurable benefit in the 28 days after an attack, on top of the systemic steroids every patient already receives. Systemic corticosteroids are powerful and broadly anti-inflammatory, and they suppress much of the same pathway. The question this trial asks is whether there is any room left above them.

The honest scientific risk. There are two, and they are different in kind.

The first is that the biology simply does not have room. If a five-day course of steroids already saturates the anti-inflammatory benefit available in the first 28 days, a targeted antibody adds nothing measurable in that window however well it works in maintenance use, and the drug’s real value lies in a longer horizon this trial does not measure. Nothing in rademikibart’s maintenance data addresses this, because no maintenance trial gives every patient systemic steroids at baseline.

The second is statistical rather than biological, and it is the larger of the two: at roughly 80 patients per arm, the trial can only detect a big effect on a binary 28-day endpoint. That is discussed with numbers in A.2’s feasibility matrix, and it is the main reason this analysis’s probability of hitting the stated primary endpoint sits below 50%.

A.2 Clinical development plan, timeline, feasibility, resourcing

Trials

Timeline specifics. First patient in 2025-08-08 [VERIFIED — CT.gov start_date]. Last patient in June 2026 — the company announced enrollment complete on 2026-06-17 and had said on the same date that enrollment was expected to complete that month [VERIFIED — CNTB press release 2026-06-17]. Last patient out follows from the protocol rather than from a disclosure: the primary endpoint reads out at Day 28 and the adverse-event window runs 56 days, so the final patient’s last visit falls in August 2026 — which is exactly what CT.gov’s completion_date of 2026-08-10 records [VERIFIED — CT.gov NCT06940141]. Database lock and topline are not disclosed as separate dates; the company gives only “September 2026”. The Readout section below models the gap explicitly rather than assuming it.

Feasibility matrix

Statistical power — the number the matrix above cannot hold in a box. This deserves its own arithmetic, because it is the single largest driver of the probability locked at the bottom of this document.

The trial randomises 160 patients, so roughly 80 per arm. The endpoint is binary: each patient either fails treatment within 28 days or does not. Published relapse rates after emergency-department treatment for acute asthma, despite systemic steroids, are 10–20% within two weeks and 25.3% within three weeks in one cited study [WEB ESTIMATE — Cochrane review CD000195 and the acute-asthma corticosteroid literature, via web search 2026-08-13]. This trial’s composite is broader than relapse alone — it also counts death, hospital admission and any treatment intensification — so a placebo failure rate of roughly 20–30% at 28 days is the reasonable expectation [UNVERIFIED — derived from the relapse benchmark above by widening for the extra composite components].

With 80 patients per arm and a 25% placebo rate, reaching two-sided p < 0.05 at 80% power requires the failure rate to fall to roughly 11% — a 56% relative reduction, 14 percentage points absolute [UNVERIFIED — standard two-proportion sample-size arithmetic on the placebo rate above]. For comparison, dupilumab’s maintenance effect on exacerbation rates is a 44–48% relative reduction over a year, and rademikibart’s own Phase 2b showed 7.5–9.3% versus 16.7% over 24 weeks in a maintenance population, which is a 44–55% relative reduction. So the effect size this trial needs is at the top of, or slightly beyond, what the class has shown in its easier setting — and it needs it in 28 days, on top of steroids.

That is not impossible. It is, however, why a reader should expect the statistical result to be harder than the biological one, and why the company’s own framing points at the secondary endpoint.

Resourcing sufficiency. Sufficient for this readout, and that is the only question that matters here. The trial is fully enrolled and study completion is passed, so the remaining spend is database lock, analysis and reporting. On cash, both readings in ../company.md C.3 — the company’s stated runway of at least one year from 2026-08-12, and this analysis’s recomputed 5.9 months from 2026-06-30 — clear a September 2026 readout with months to spare, so the answer does not depend on which reading is right. Beyond the readout the two readings diverge materially, and the Phase 3 the company intends to negotiate with the FDA afterwards is not funded on either of them: a global Phase 3 in this setting would cost multiples of the $31.5M on hand. That is a financing question for after the event, and it is priced into the run-up section’s financing driver rather than into this row.

A.3 Target label, target product profile, strategic questions, designations

A.3a Target label. Rademikibart as an add-on to standard of care for the treatment of an acute exacerbation in adults and adolescents aged 12 and over with asthma and type 2 inflammation (blood eosinophils ≥300 cells/µL), administered as a single dose in the urgent-care setting [VERIFIED — CT.gov NCT06940141 title and eligibility criteria].

A.3b Target product profile

A.3c Strategic Go/No-Go questions. The set that matches this asset’s next decision is pre-Phase-III: the company’s stated plan after topline is to “meet with the FDA to gain alignment on a Phase 3 program” [VERIFIED — exhibit 99.1 to CNTB Form 8-K, 2026-08-12].

Pre-Phase-III (Go-to-Phase-III / registration):

A.3d Regulatory designations. None disclosed for this program. No Fast Track, Breakthrough Therapy, Orphan Drug or other designation appears in the company’s press feed, in the registry record, or in the filings read at the company tier [VERIFIED — BPIQ fetch_company_press_releases 2026-08-13, 152 items; CT.gov NCT06940141; EDGAR submissions]. The absence is unsurprising rather than alarming: asthma is not a rare disease, so Orphan Drug does not apply, and the designations that could apply generally follow a positive randomised readout rather than precede one.

Two regulatory interactions are on record and are not designations:

• Positive Type C meeting with the FDA, announced 2025-04-01. A Type C meeting is a general guidance meeting between a sponsor and the FDA. It grants nothing; it means the agency gave feedback the company considered favourable [VERIFIED — CNTB press release 2025-04-01].
• Favourable FDA feedback on a potential Phase 3 programme, recorded 2024-09-05 in the third-party feed against the atopic-dermatitis row [VERIFIED — BPIQ fetch_company_drugs note, id 16030]. Different indication; noted for completeness.

A.4 Target product profile valuation matrix

The provider row is the underrated one. A drug for the acute setting has to survive contact with an emergency department at three in the morning. The eosinophil count is already on the chart, so patient selection costs nothing — but a subcutaneous biologic is not something an emergency department stocks or gives. That the company started an intravenous bridging study in August 2026, before this readout, suggests it understands the constraint. It also means a positive readout here is not immediately commercialisable in the form tested.

A.5 Evidence quality and endpoints

Endpoints

[VERIFIED — CT.gov get_trial_details NCT06940141, 2026-08-13, all endpoints and time frames as recorded]

A.5b Key opinion leaders.

Both tables below are empty, and each empty table is a finding rather than a blank section. What was searched, and what came back, is stated in full so a reader can judge the absence.

Panel as of. 2026-08-13 — the date the investigator and independent-voice searches below were run.

Investigators

No investigator is recorded, because ClinicalTrials.gov carries none for this trial. get_trial_details on NCT06940141 returns all 58 sites with contacts: null and no overall-official record at all — normal for a trial that has finished recruiting, since contact details are removed once enrollment closes. search_investigators was then run twice as a cross-check: the first attempt returned 0 of 0 on a bad parameter combination, and the second, on the condition alone, returned 21 named investigators across 40 analysed trials, none of them on a Connect Biopharma trial. So the mechanism works and the record is genuinely empty [VERIFIED — CT.gov get_trial_details and search_investigators, 2026-08-13].

Independent voices

No independent voice is recorded, and the reason is a connector gap rather than an absence of commentators. Three independent syntheses covering rademikibart exist and were retrieved — the Australasian Journal of Dermatology meta-analysis (PMID 42117199), the Cureus Bayesian network meta-analysis (PMID 41884322) and the Pharmacoepidemiology and Drug Safety network meta-analysis (PMID 41346302) — and the first of them takes a substantive position on the molecule. But PubMed’s get_article_metadata returned the authors field as nulls on all 14 articles, so not one of those commentators could be named. Rule 40 requires a name, a date, a source and a conflict disclosure, or the view is not recorded; without a name, the view is not recorded. It is not attributed to “the authors of” anything, because that is exactly the unnamed, undated, unconflict-checked attribution rule 40 exists to forbid.

Two further notes on what was deliberately not put in this table. First, the authors of rademikibart’s own trial publications would not belong here even if they could be named: they are the sponsor’s investigators, which is the other table. Second, the seven named sell-side analysts whose price targets appear under Market and timing below are not independent clinical voices, and recording them as such would repeat the exact v3.0.0 rule violation that framework/01-rules.md rule 40 carries as its own worked failure. They are used only as the published-analyst-target anchor that rule 30 admits, and nowhere else.

Judgement

Dissent

Empty, and correctly so: endpoint_supported is UNKNOWN rather than SUPPORTIVE_CONTESTED, so there is no claimed support for anyone to contest. Inventing a disagreement nobody has voiced would be worse than an empty table.

B. Commercial assessment

B.0 Current treatment algorithm

What a patient having an asthma attack receives today, line by line, and exactly where this drug would fit:

1. At home, first. A short-acting reliever inhaler, or a combination inhaler used as a reliever. If that controls the attack, the patient never reaches this trial’s population.
2. At the urgent visit — where this drug would be given. Repeated nebulised or inhaled bronchodilators, oxygen if needed, and systemic corticosteroids — tablets or injection. This is the standard of care and it has not fundamentally changed in decades. Every patient in Seabreeze STAT Asthma receives it, and rademikibart is added on top of it rather than replacing any part of it.
3. On discharge. A short course of oral steroids, typically about five days, plus a review of the patient’s daily controller inhalers. Continuing steroids after discharge measurably reduces relapse, and the benefit lasts about three weeks [WEB ESTIMATE — Cochrane review CD000195 via web search, 2026-08-13].
4. The 28 days that follow — the gap this drug is aiming at. Despite all of the above, 10–20% of patients relapse within two weeks of emergency-department discharge, and 25.3% within three weeks in one cited study [WEB ESTIMATE — Cochrane review CD000195 and the acute-asthma corticosteroid literature via web search, 2026-08-13]. That relapsing fifth to quarter is the entire commercial opportunity, and there is no approved drug that addresses it.
5. Weeks to months later, for a subset. Patients with frequent attacks may be referred for a maintenance biologic — dupilumab, or an anti-interleukin-5 agent. This is where the class lives today, and it is a different decision made at a different time by a different doctor. Dupilumab’s own label states it is not for the relief of acute bronchospasm.

Where rademikibart would fit: step 2, as a single add-on dose, for the roughly half of patients whose attack is driven by type 2 inflammation. It does not displace anything. It is additive to steroids, which is both its commercial advantage — no prescriber has to stop doing what they already do — and its clinical risk, because it has to prove benefit on top of a therapy that already works (A.1).

B.1 Competitive landscape and positioning

Where this asset wins. In a setting nobody else is in. The competitive landscape for IL-4Rα blockade in maintenance asthma is crowded and dupilumab has a decade of head start, eight approvals and a brand every pulmonologist knows. Rademikibart cannot win there on a 2.2× binding advantage. What it can do is define a new use — a single dose at the moment of the attack — where the incumbent is not merely absent but labelled out.

The single fact the thesis rests on. That rademikibart’s lung-function effect arrives fast enough to matter inside 28 days. Everything else — the tighter binding, the crystal structure, the absent hypereosinophilia, the routine biomarker — is supporting evidence for a drug that is already known to work in this disease over months. Only speed distinguishes an acute drug from a maintenance drug, and the evidence for speed is one Phase 1 study reporting effects from 15 minutes and a Week 1 signal in the Phase 2b. This readout is the first randomised, blinded, placebo-controlled test of that single fact.

Calibration reference from 03b-program-spec.md, not a claim about this asset: oncology first-movers in novel mechanisms have shown roughly 3.2× higher peak-sales potential but about 1.8× higher development risk.

B.2 Addressable market

Launch markets. The United States first — 22 of the trial’s 58 sites are American, the company is headquartered in San Diego, and it has run its regulatory interactions with the FDA. Then the European Union and the United Kingdom, with UK sites already in the trial. Greater China is not Connect’s market at all: those rights are exclusively licensed to Simcere, and Connect’s economics there are milestones and royalties [VERIFIED — CT.gov NCT06940141 locations; exhibit 99.1 to CNTB Form 8-K, 2026-08-12].

B.3 Value and feasibility

B.3a Expected peak sales.

First, the third-party figure and why this document does not use it. The company’s own market research forecasts peak sales “upwards of $5 billion” globally, and Northland’s Carl Byrnes initiated coverage seeing peak-sales potential exceeding $5 billion [WEB ESTIMATE — Northland initiation and company commentary via web search, 2026-08-13]. That figure is not passed through here, for a specific reason rather than general scepticism: it is explicitly for “acute and chronic asthma/COPD indications” — that is, it includes rademikibart competing head-to-head with dupilumab in the maintenance market, which this readout does not address and which the company is not currently running a trial in. Presenting a $5 billion number against a readout about single-dose acute treatment would answer a different question than the one asked. The build below is for the acute asthma exacerbation indication only — the indication this catalyst is about.

Every input is tagged. Three of the four are unverified, which is why every row is a range and why none of them should be read as a forecast.

Conditional on approval, and this excludes: the COPD acute indication (a separate readout, on a separate trial, and a separate market roughly comparable in size); any maintenance asthma or COPD indication (no trial is running); Greater China entirely (licensed to Simcere — Connect’s economics there are up to approximately $99M of remaining milestones plus tiered royalties to low double-digit percentages on net sales [VERIFIED — exhibit 99.1 to CNTB Form 8-K, 2026-08-12]); and the existing atopic-dermatitis franchise.

Which input is least defensible: the price. Patient numbers can at least be bounded by epidemiology, and penetration by analogy to other injectables. There is no drug priced for a single acute dose in this setting anywhere in the world, so the $2,000–$6,000 range is an inference from a maintenance list price and an avoided-admission cost, not a benchmark. A reader who disagrees with the price should scale every row proportionally.

What this means against the market value. Enterprise value is $115.24M (../company.md C.4). The base case of roughly $0.3B in annual peak sales for one indication is about 2.7× that enterprise value, and the low case of $0.05B is about 0.4×. For an asset this early those multiples are ordinary rather than remarkable — which is the honest reading: the stock is not obviously mispriced against a risked view of this indication alone. What makes the setup interesting is the optionality stacked behind it (the COPD readout in the same month, the intravenous presentation, the China royalty stream), not a screaming valuation gap on this program.

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Only partly, and the gap is specific. The plan tests whether a single subcutaneous dose reduces 28-day treatment failure — which is the efficacy claim and the payer claim in one endpoint, so those are well served. It does not test the presentation the target profile’s premium row actually calls for: an intravenous push that fits an emergency-department workflow. That is a separate 40-patient open-label study (Seabreeze STAT IV) reading out around December 2026, and it is single-arm, so it can bridge pharmacokinetics but cannot establish efficacy of the intravenous form.
2. Will the identified risks affect the target product profile? The two that would are the statistical-power risk and the steroid-background risk, and they affect the same row: efficacy. If the primary endpoint misses on power while FEV₁ separates cleanly, the profile survives but needs a larger Phase 3 with a different primary endpoint — which is what the company has all but said the topline is for. If FEV₁ does not separate at Week 1, the rapid-onset claim that differentiates this asset from dupilumab fails, and the profile does not survive that.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? Yes, but on a much thinner basis. Strip out the acute indication and rademikibart is a next-generation IL-4Rα antibody with a 2.2× binding advantage, an apparent absence of the class’s eosinophil rise, a positive maintenance asthma Phase 2b, positive Phase 2 and Phase 3 atopic-dermatitis data and a China partner — a real asset, competing against an entrenched incumbent on modest margins, and worth far less than a first-in-setting drug.

Readout

Sources

The modelled estimate. CT.gov’s primary_completion_date for NCT06940141 is 2026-07-17. Add the default lag to database lock and analysis, which 01-rules.md rule 34 sets at two to four months: 2026-09-17 to 2026-11-17. The tag is [UNVERIFIED — modelled, default lag] rather than benchmarked, because data/benchmarks/readout-lag.json holds zero observations — no comparable trial’s lag has been confirmed in this repository yet, so there is nothing to benchmark against [VERIFIED — data/benchmarks/readout-lag.json read 2026-08-13, observations array empty].

Two things are worth noting about how this modelled range sits against the company’s guidance. It starts later than the company’s own month opens, and it runs two months past where the company expects to be finished. That is what a generic default lag does to a trial with an unusually short endpoint: the default is calibrated on trials where database lock follows a long, complex follow-up, whereas here the primary endpoint is a 28-day binary composite in 160 patients and the safety window closed on 2026-08-10. This trial should lock and analyse faster than a default assumes. The window below therefore leans on the company’s guidance for its early edge and borrows from the modelled range only for its tail — and says so, rather than averaging the two.

Window

Basis. The earliest edge is the first day of the month the company itself named eight days ago, with the trial’s own safety follow-up already closed on 2026-08-10, so nothing operational stands between now and a first-week announcement. The likeliest date is mid-September, where three sources converge from different directions: the third-party feed’s synthesized 2026-09-15, the modelled range’s own opening at 2026-09-17, and the middle of the company’s stated month. The latest edge sits one month past the guided month rather than at the modelled range’s 2026-11-17 tail — because the default lag is calibrated on trials with far longer follow-up than a 28-day endpoint whose safety window has already closed, while this program’s two prior slips are real and argue against taking the company’s month as a hard ceiling. Precision is MONTH because a source names a month and none names a day. Confidence is MEDIUM and not HIGH: enrollment is complete, the endpoint is short and fixed, and the company reaffirmed eight days ago — all of which argue up — but the date has slipped twice and the answer is still a month rather than a day.

Disagreement. CONSISTENT. Three sources point at September 2026 from different directions and none contradicts another: the company says “September 2026”, the feed’s synthesized day falls inside it, and the modelled range overlaps it and extends past it. The ctgov row is not in disagreement with any of them because it answers a different question — when the last patient finished the primary assessment, not when the result is announced. There is no source pointing at a materially different date, so nothing here is averaged and nothing is resolved by picking one.

Date slippage. Two slips across eight dated statements. Parsed from the BPIQ note field for id 13433 and from the company’s own releases, oldest first.

What the pattern says. Both slips happened while enrollment was still the constraint, and the second one came with the announcement that the constraint had been removed. Six of the eight statements are reiterations. This is not a programme that has been quietly pushing a date every quarter; it is one that missed its original guidance by roughly three months for a reason it named (recruitment in an acute setting is hard) and has held the new date since. That is why confidence is MEDIUM rather than LOW.

Attribution

Status. CONTAMINATED

Computed by lib/clustering.mjs’s attributionFor("CNTB", companyRecord, programs, 6, 13433) over this ticker’s full pipeline table on 2026-08-13, and transcribed rather than judged by eye.

Conflicts

Note. Required, because the status is CONTAMINATED.

The conflicting row is the Seabreeze STAT COPD study (NCT06940154, n=159), and it is not a coincidence of two placeholders touching. The gap is zero days — the ranges overlap outright — and the conflict is confirmed: true because at least one side rests on a disclosed date: this program’s own readout.window carries MONTH precision.

The contamination is also not merely arithmetic. The company’s own words, in a filed exhibit, are that it expects “to report topline results from both Phase 2 Seabreeze STAT studies in September 2026”, and its chief executive frames the pair as one event: “we are approaching a key inflection point with randomized, controlled data of rademikibart in patients experiencing an acute exacerbation” [VERIFIED — exhibit 99.1 to CNTB Form 8-K, 2026-08-12]. Both studies share a sponsor, a molecule, a mechanism, a primary endpoint definition, a key secondary endpoint, an eosinophil threshold and many of the same clinical sites. It is entirely possible the two toplines are announced in the same press release.

The consequence for the scenario range and the stock call. Neither can be read as belonging to the asthma result alone, and this document does not pretend otherwise. Concretely:

• A positive asthma result paired with a COPD miss would not deliver the positive scenario range below. COPD is the larger commercial prize and a public failure in it would substantially offset a win in asthma.
• A miss in asthma paired with a positive COPD result would not deliver the miss range either.
• The correlation is high but not perfect, and that is the useful nuance. Same drug, same mechanism, same endpoint definition — so the two results are far from independent, and the most likely outcomes are both-positive or both-negative. But the diseases differ in ways that could separate them: COPD patients are older, are by definition smokers or ex-smokers (whom this asthma trial excludes above 10 pack-years), and type 2 inflammation explains a smaller share of COPD than of asthma.

The scenario ranges below are therefore written as ranges for the two-study package, weighted toward the asthma result because it is the larger of the two trials and the indication with the stronger prior evidence for this molecule. They are not clean per-program ranges, and the low ends of both are set wider than an attributable single-program readout would need, precisely because of the split-outcome cases above.

Market and timing for this event

• Plain takeaway. A 19-day wait into a binary readout on the company’s dominant program, with the stock at 42.5% of its 52-week range, short interest nearly doubled since June, one director buying in the open market at prices above today’s, and an options chain that cannot price anything. The payoff is asymmetric upward — roughly +117% on the positive midpoint against −45% on the miss midpoint — which is why the stock call and the outcome call point different ways.
• Months to this catalyst. 0.6 months — 19 days from 2026-08-13 to readout.window.earliest of 2026-09-01. Measured from the earliest edge of the readout window and never from catalyst_date (rule 23). Said plainly, as the template requires: readout.precision is MONTH, not DAY, so this is a distance to the opening of a month-wide window, not to a known event date. The likeliest date is 33 days out and the latest edge is 79 days out.
• Expected move around this event. ../company.md C.6 records the chain as unusable, so this is a bracket and not a point estimate: roughly ±35% to ±65%, inferred from the 2026-09-18 $2.50 call quoted at $0.35–$0.60 on a $2.33 stock. Treated as a sanity check on the scenario ranges below and never as an input to them. Note also that the 2026-09-18 expiry does not cover the latest edge of the readout window; the first expiry that does is 2026-12-18.
• Nearest comparable past reaction. The closest analogue in ../company.md C.7 is 2025-06-13, +13.21% — rademikibart asthma data at EAACI 2025, with a clean same-day press feed. Why it is only partly comparable: it was a presentation of data already known to be positive, not a blind binary readout, so it measures the market’s appetite for good news rather than its reaction to news that could go either way. The second-closest, 2024-05-22 at +11.73%, has the same limitation. Nothing in C.7 is a comparable negative reaction at all — the one large decline, −19.89% on 2026-03-30, was bundled with a $20.2M placement — so the downside side of this ticker’s record is unpopulated rather than reassuring.
• Materiality. Dominant, but shared — the exact wording recorded for this program in ../company.md C.2. Dominant because a clear miss reprices a company whose other eight pipeline rows comprise two failed programs, three effectively shelved ones, two stale duplicate catalyst flags and one partnered franchise already read out. Shared because the COPD study reads out in the same month. The stock-direction call below is made consistent with both halves (rule 27): direction up, confidence low, and the note carries the attribution caveat.
• Date slippage. Two slips across eight dated statements — see Readout above, readout.slips. Both occurred while enrollment was the binding constraint, which it no longer is.

Spot. $2.33, read 2026-08-13, cited from ../company.md C.1 [VERIFIED — BPIQ fetch_company_info 2026-08-13].

Scenario prices

Expected value. $2.71, +16.3% against spot of $2.33.

Method: the 38% probability from the outcome call applied to the midpoint of each range — 0.38 × $5.05 + 0.62 × $1.275 = $2.7095, rounded to $2.71, then divided by $2.33 and less one.

This is arithmetic, not advice, and it is not a price target. It is what the probability and the two ranges imply and nothing more. Note where the positive number comes from: not from thinking the trial will succeed — the same arithmetic says it probably will not — but from the payoff being lopsided. The positive midpoint is +117% against spot while the miss midpoint is −45%, so a 38% chance of the first outweighs a 62% chance of the second.

Run-up

A run-up call IS locked on this program. Rule 39 withholds one only where the date_confidence driver floors at zero, which happens exactly when readout.precision is UNKNOWN. Here precision is MONTH and confidence is MEDIUM, so the driver scores 48 — above lib/runup.mjs’s untradeable threshold of 30, and well above the floor. There is a date to time an entry and an exit against.

Entry and exit

exit is null on the locked record, and would be even with more data: resolveExit needs trading bars before readout.window.earliest, which is a future date. Separately, and worth carrying forward, there is no committed price cache for this ticker at all — data/prices/CNTB.json does not exist (../company.md C.8 row 8) — so this run-up call cannot be settled until that cache exists, because 05-prediction-protocol.md requires settlement to read the committed cache and never a fresh fetch.

Predicted move and peak

Priority score drivers

Five of the seven are transcribed from lib/runup.mjs’s scoreDriver, run on 2026-08-13; two are the analyst’s own judgement, read from this document.

Priority score. Transcribed from lib/runup.mjs’s priorityScore, never hand-computed.

Priority score 23 · formula_version 1.0.0

What a 23 means, plainly. Low. Six of the seven drivers are middling-to-strong — unmet need and value uplift at 85, room left to run at 58, squeeze at 59 — and the score is nonetheless 23, because financing_clustering_risk is at its maximum and date_confidence of 48 gates everything else multiplicatively. The single thing sinking this program’s run-up ranking is the simultaneous COPD readout, not anything about the asthma trade itself. A reader who thought the two toplines would land months apart would be looking at a materially higher score, and that is the right way round: the formula is saying you cannot cleanly own this event on its own.

Settlement. Null at lock time. Settled only on an explicit user request, from a confirmed primary source, against the committed price cache — and, as noted above, not settleable at all on this ticker until data/prices/CNTB.json exists.

Verdict

What I would do. Watch — with one specific exception, stated below.

Why. The science here is better than the trial design. Rademikibart is genuinely and measurably differentiated from the one approved drug in its class, it works fast, and it is aiming at a setting where nothing is approved and the incumbent is labelled out. But the trial that reads out in September has roughly 80 patients per arm on a binary 28-day endpoint, needs something close to a halving of the placebo failure rate to reach significance, and has to achieve that on top of the systemic steroids every patient receives — and the company’s own framing, that the topline “should help determine a potential Phase 3 endpoint and sample size requirements”, reads like a study built to inform a Phase 3 rather than to win one. On top of that, the COPD study reads out in the same month, so even a correct read on the asthma biology does not give you a clean trade. The exception: the run-up leg is a separate, separately scored call and it is pre-registered above, because a 19-day hold into a well-telegraphed binary with doubled short interest and a call-skewed chain is a different proposition from holding through the event. Holding through the event is what I would not do.

What would change this. Two observables, in opposite directions.

• To the upside: disclosure of the placebo-arm treatment-failure rate assumption, or of the powering assumption, showing the trial was designed against an effect the drug has already shown. The power arithmetic is the main thing holding the probability below 50%, and it is arithmetic done from outside on an assumed placebo rate. A protocol or a statistical-analysis-plan disclosure that makes the assumed effect explicit would move the number materially, either way.
• To the downside: a raise before the readout. Both runway readings clear September, so a placement now would say something about management’s own confidence that the arithmetic cannot.

What to watch. Dated, checkable items between now and the event:

• 2026-09-01 onward — the topline itself. Watch the key secondary before the primary: absolute change from baseline in post-bronchodilator FEV₁ at Week 1. That is where this drug’s evidence is strongest and where a rapid-onset claim either appears or does not. The Day 3 FEV₁ timepoint is the sharper version of the same test.
• On the same day — whether the COPD topline is in the same release, and whether the two agree. A split result is the case neither scenario range above describes cleanly.
• On the same day — the placebo-arm treatment-failure rate. If it lands below about 20%, the trial was fighting a harder fight than this analysis assumed and a miss on the primary carries less information about the drug.
• Any date before the readout — an 8-K or a 424B5 indicating a financing. See “What would change this” above.
• Any date before the readout — further Form 4s from James Huang or any other director. Two open-market purchases at $3.45 and $2.4774 are on record; a third, or a first sale, would be informative.
• Late 2026 — the FDA meeting on a Phase 3 programme, which the company says follows topline. A positive Phase 2 whose endpoint the agency will not accept for a pivotal is a materially worse outcome than it looks.
• Around 2026-12 — Seabreeze STAT IV primary completion (NCT07705737). The intravenous presentation is what makes this drug usable in an emergency department at all.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): miss — probability 38%, band 26–52% [UNVERIFIED — modelled]. The probability that the settlement definition below is met. The point sits below 50 because statistical power at roughly 80 patients per arm is the binding constraint: reaching two-sided p < 0.05 needs the treatment-failure rate to fall from an expected 20–30% to around 11%, at or beyond the top of what IL-4Rα blockade has achieved in its easier maintenance setting, and it has to do so on top of the systemic steroids every patient receives. The band is wide and crosses 50 because a large effect is genuinely plausible — the trial enriches for the eosinophil-high patients in whom this molecule showed +420 mL of placebo-adjusted lung-function gain, the drug acts within days rather than weeks, and the data monitoring committee saw no reason to enlarge the study at its April 2026 interim look. No third-party probability on this event was found, so there is no published figure to position against; the seven sell-side price targets are opinions about value, not about this endpoint.
• Stock-direction (which way do the shares move?): up — confidence low — window 2026-08-13 to 2026-11-14, basis: the window opens at the lock date and runs two weeks past readout.window.latest of 2026-10-31, so the call still resolves if the readout slips a month as it has twice before. Why this points the other way from the outcome call, and why that is not a contradiction: the two are separate calls on separate evidence (05-prediction-protocol.md). The payoff is lopsided — the positive midpoint is +117% against spot and the miss midpoint is −45% — so a below-even probability still produces a positive expected value. And the market will read the Week 1 lung-function secondary and the Phase 3 path, not only the strict 28-day composite. Confidence is low, not medium, for two stated reasons: ../company.md C.2 records this program’s materiality as dominant, but shared, and Attribution above is CONTAMINATED — the COPD readout lands in the same month, so this move is not attributable to the asthma result alone.
• Scenario prices: positive $3.60–$6.50 · miss $0.95–$1.60 (from the table above; each side built on at least two named, individually tagged anchors per rule 30)
• Expected value: $2.71, +16.3% against spot $2.33 (arithmetic, not advice)
• Run-up: entry $2.33 on 2026-08-13, exit rule T-2 trading days before readout.window.earliest — predicted move +8% to +30%, predicted peak +15% to +45% around 2026-09 — priority score 23, formula_version 1.0.0
• Settles on the company’s own topline announcement of the Phase 2 Seabreeze STAT Asthma trial (NCT06940141), or the ClinicalTrials.gov results record for that trial, whichever is available first. Positive is defined as: rademikibart plus standard of care shows a statistically significant reduction (two-sided p < 0.05) in the treatment-failure rate within 28 days after randomisation versus placebo plus standard of care, in the trial’s primary analysis population, as reported by the company or the registry. Treatment failure is as the registry defines it: death from any cause, admission or re-admission to hospital for asthma, an emergency-department revisit or unscheduled medical visit for worsening asthma, or the necessity to intensify pharmacologic treatment including a second course of systemic steroids. Explicitly NOT positive under this definition: a numerical reduction that does not reach p < 0.05; a win on the key secondary (post-bronchodilator FEV₁ at Week 1) without one on the primary; or a positive result in a subgroup only.
• Locked: yes · Settled: no

Program data-quality flags

1. PubMed get_article_metadata returned the authors field as nulls on all 14 articles. Titles, journals, DOIs, dates and full abstracts all came back correctly. This is the single most consequential connector gap in this analysis: it is why A.5b’s independent-voices table is empty and its judgement reads UNKNOWN, and why A.5’s publication-quality row cannot assess author independence. Not previously documented in 02-connectors.md.
2. Open Targets search_entities BLOCKED — Rate limit exceeded for client: global on all three attempts. The standing platform throttle 02-connectors.md documents, reproduced. Cost: no independent human-genetics validation of IL-4Rα, so that leg of A.5’s target-validation row is [UNVERIFIED].
3. ChEMBL compound_search BLOCKED — Internal error: Error calling tool 'compound_search': HTTP error in tool compound_search: 500 from upstream API on all three attempts, twice on rademikibart and once on CBP-201. A server-side 500, not a rate limit, and not previously documented in 02-connectors.md — the only ChEMBL note there concerns unreliable indication search, not a hard failure of compound_search itself. Cost is small: the molecule is an antibody, whose selectivity question is answered better by the published binding and structural work.
4. A source conflict left open rather than smoothed: enrollment in the COPD study. The company states 159 participants; CT.gov’s record for NCT06940154 says 160. Both are recorded and neither is picked. It does not affect this program.
5. readout.sources[].congress is null, and the reason is recorded rather than inferred. Two independent sufficient reasons — data/congresses.json contains no respiratory meeting at all, and the company has named no congress for this topline. A respiratory congress in the window is plausible given this company’s presentation history, but matching on therapeutic area alone is a guess, so nothing was filled in.
6. data/benchmarks/readout-lag.json is empty, so the modelled readout estimate uses the stated default of primary completion plus two to four months and is tagged [UNVERIFIED — modelled, default lag] rather than benchmarked. On a trial with a 28-day primary endpoint whose safety window is already closed, that default is probably too conservative, and the Basis paragraph under Window says so rather than silently adopting it.
7. The trial’s own eligibility criteria carry two different eosinophil thresholds, and both are real rather than an error: a historical threshold of ≥250 cells/µL (or FeNO ≥25 ppb) measured when the patient was stable, and ≥300 cells/µL measured at the index acute exacerbation itself. This document uses ≥300, the enrolling threshold at the qualifying visit. Recorded because a reader comparing against the company’s own “≥300 cells/µL” phrasing would otherwise find the registry text confusing.
8. CT.gov carries no named investigator for NCT06940141. All 58 sites return contacts: null and there is no overall-official record. Normal for a trial that has closed recruitment, and confirmed as a real empty rather than a tool failure by a second search_investigators call that returned 21 investigators on other trials in the same condition. See A.5b.
9. The registry says Seabreeze STAT IV is NOT_YET_RECRUITING while the company says it initiated the study in August 2026. The registry lags. Recorded as a disagreement; it does not affect this program’s own readout.
10. No patent or exclusivity information could be obtained. The dedicated web search explicitly returned none for rademikibart’s composition-of-matter claims or expiry. B.1’s IP row and B.2’s exclusivity row are [UNVERIFIED] with no assumption substituted, and B.4’s IP row records the unknown as a risk rather than dismissing it.
11. Two run-up inputs depart from the standard computation because there is no price cache for this ticker, and both departures are stated where they are used: priced_in’s 52-week range comes from BPIQ under 03a-company-spec.md C.4’s stated exception rather than from lib/prices.mjs’s range52w, and the run-up call cannot be settled at all until data/prices/CNTB.json exists. See ../company.md C.8 row 8.