IRON / disc-3405-polycythemia-vera — DISC-3405 for polycythemia vera

Program analysis · bpiq_drug_id 18007 · prepared 2026-08 · USD · framework v5.11.0 · 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].

This is a full refresh, replacing the 2026-08-13 analysis and the 2026-08-15 re-lock. Both tiers of the sweep were re-run in full on 2026-08-20; the delta window reported throughout is 2026-08-13 → 2026-08-20. The re-lock stamp the previous version carried is gone because one sweep now produces both facts again (CHANGELOG.md v5.10.3).

Glossary

Executive summary

• What it is (one sentence): DISC-3405 is an antibody, injected under the skin, that blocks a liver protein called TMPRSS6 so the patient’s own body makes more hepcidin, which locks iron away and starves the runaway red-cell production that defines polycythemia vera.
• The event and when (as disclosed): Initial data from the Phase 2 RESTORE-PV trial, guided to “Q3 2026” — a quarter, not a month and not a day. The company said on 2026-07-30 that enrolment is complete and the data are “coming ahead of schedule in Q3” [VERIFIED — exhibit 99.1 to Form 8-K filed 2026-07-30, re-read in full 2026-08-20]. Nothing has printed in the twenty-one days since, and the company has issued no release of any kind in that period.
• The main reason it could work: The pharmacology is already demonstrated in humans and the target is already clinically validated by others. DISC-3405’s published Phase 1 in 56 healthy volunteers raised hepcidin and lowered serum iron, transferrin saturation, reticulocyte haemoglobin, haemoglobin and haematocrit — exactly the chain of effects polycythemia vera needs [VERIFIED — J Clin Pharmacol 2026;66(5):e70199, via PubMed]. And in the last twelve months two other drugs working on the same hepcidin pathway have succeeded in this indication in placebo-controlled trials.
• The main risk: Not that the drug fails to work — that it works and arrives third. Rusfertide holds an FDA decision in the same quarter as this readout, still undecided as of 2026-08-20, and divesiran, which silences the identical target, posted an 88% versus 19% placebo-controlled Phase 2 result on 2026-08-10 and has now set a Phase 3 start for the first half of 2027. RESTORE-PV is open-label, single-arm, 60 patients, and its own registered primary endpoint is safety, not efficacy — so a good number here will be read against two better-controlled numbers that already exist.
• What it means for the stock: Less than its science deserves, and slightly less than a week ago. This is the second-most-important program on a ticker whose equity trades on bitopertin, whose Phase 3 APOLLO readout lands one quarter later. The options market prices the December expiry at roughly double the implied volatility of the September one (../company.md C.6), and the September expected move has itself come in from 12.0% to 10.4% in seven days. In ten years of catalyst history only bitopertin has ever moved these shares more than 10% — with one exception, and that exception is this molecule.

0. Program-tier coverage — CLEARED

One row per mandatory tool in the program-tier table in framework/02-connectors.md, in the order that file lists them, followed by its optional rows. Company-tier coverage is in ../company.md C.0. Every row below was re-run in full on 2026-08-20; a refresh that reuses the program-tier sweep is a document edit, not an analysis.

No mandatory row reads NOT CALLED, and this sweep has no BLOCKED mandatory row at all — an improvement on the previous version, which carried Open Targets as its one block. The program is CLEARED and the prediction can be locked (02-connectors.md § The gate).

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

What the drug is. DISC-3405 is a recombinant humanized monoclonal antibody — a laboratory-made protein, engineered to look like a human antibody so the immune system tolerates it, that sticks to one specific target and blocks it. It is given as an injection under the skin. Its elimination half-life is about 11 days at clinically relevant doses, so dosing is measured in weeks rather than days [VERIFIED — J Clin Pharmacol 2026;66(5):e70199, [DOI 10.1002/jcph.70199](https://doi.org/10.1002/jcph.70199), via PubMed]. It was originally Mabwell Therapeutics’ MWTX-003 and was in-licensed by Disc Medicine in January 2023 [VERIFIED — Disc Medicine FY2025 Form 10-K, filed 2026-02-26].

How it works, in order. In polycythemia vera a single mutation (usually JAK2 V617F) leaves the bone marrow making red blood cells without the usual brake. The haematocrit climbs above 45%, the blood thickens, and the risk of clots and strokes rises. The standard answer is to remove blood repeatedly, which works but strips out iron and leaves patients chronically iron-deficient and tired.

Red-cell production needs iron. The body controls how much iron is available with a liver hormone called hepcidin: high hepcidin closes ferroportin, the only exit door for iron, so iron stays locked inside gut and spleen cells and never reaches the marrow. In polycythemia vera hepcidin is abnormally low, which is part of why production runs unchecked [VERIFIED — Blood 2026;147(12):1278-1288, [DOI 10.1182/blood.2025028643](https://doi.org/10.1182/blood.2025028643), via PubMed: "Because hepcidin levels are relatively low in patients with PV, hepcidin agonists ... are undergoing clinical development to control PV-associated erythrocytosis"].

TMPRSS6 is the protein that keeps hepcidin low. It sits on the liver-cell surface and cuts up hemojuvelin, the helper protein that switches hepcidin production on. DISC-3405 blocks TMPRSS6. Hemojuvelin survives, the BMP/SMAD signal stays switched on, the liver makes more hepcidin, iron is locked away, less iron reaches the marrow, and red-cell production slows for want of raw material. The intended result is a haematocrit held below 45% without taking blood out of the patient — what the field calls a “chemical phlebotomy”.

How well is the target validated? Strongly, and by three independent routes.

1. Human genetics. People born with loss-of-function mutations in TMPRSS6 have a recognised inherited condition, iron-refractory iron-deficiency anaemia, in which hepcidin is permanently high and iron permanently low. That is the human phenotype of the exact intervention DISC-3405 makes, running for a lifetime [UNVERIFIED — established in the clinical genetics literature and consistent with the mechanism described in the tagged sources above]. This sweep narrowed why it is still unverified rather than removing the gap. Open Targets’ search layer now answers and resolves TMPRSS6 to ENSG00000187045 and polycythemia vera to MONDO_0009891, but its GraphQL layer refused the association query three times with Rate limit exceeded for client: global (section 0). Resolving an identifier is not evidence of an association, so the tag stands.
2. Animal proof of concept in this exact disease. Hepcidin agonists normalised haematocrit and reduced splenomegaly in mice carrying the orthologous JAK2 mutation that causes human polycythemia vera [VERIFIED — Blood 2016;128(2):265-76, [DOI 10.1182/blood-2015-10-676742](https://doi.org/10.1182/blood-2015-10-676742), via PubMed].
3. Clinical validation of the pathway in patients, by competitors. Rusfertide, a hepcidin mimetic, met its Phase 3 primary endpoint in polycythemia vera. Divesiran, an siRNA that silences the same TMPRSS6 target DISC-3405 blocks, met its Phase 2 primary endpoint on 2026-08-10 and is going to Phase 3 [VERIFIED — see B.1 for both, with numbers]. The pathway is no longer a hypothesis in this indication.

The exact scientific step the next readout must prove. Not that raising hepcidin restricts iron — DISC-3405’s own Phase 1 already showed that in humans, with dose-dependent rises in hepcidin-25 and falls in serum iron, transferrin saturation, reticulocyte haemoglobin, haemoglobin and haematocrit in 56 healthy volunteers [VERIFIED — J Clin Pharmacol 2026;66(5):e70199]. What RESTORE-PV must prove is the step after that: that in actual polycythemia vera patients, whose marrow is being driven by a mutation rather than by normal physiology, the iron restriction is strong enough and durable enough to keep haematocrit under 45% and stop the phlebotomies — and that it can be done without pushing patients into the symptomatic iron deficiency that phlebotomy itself causes.

The honest scientific risk. Two, and neither is about whether the mechanism works.

The first is dose control with a long-acting antibody. DISC-3405’s clinically relevant half-life is about 11 days [VERIFIED — Phase 1 publication]. That is an advantage for convenience and a liability for titration: if a patient’s haematocrit is driven too low, or iron restriction becomes symptomatic, the drug cannot be withdrawn quickly. Rusfertide is a short-acting peptide and divesiran is dosed every 6 to 12 weeks with a titratable effect; DISC-3405 is the least reversible of the three. RESTORE-PV’s within-participant dose escalation across two dose levels is the design response to exactly this problem, and whether it worked is part of what the readout answers.

The second is immunogenicity — the risk that patients form antibodies against the antibody, blunting it over time. The Phase 1 measured this in healthy volunteers over weeks; polycythemia vera is a lifelong disease and the relevant question is years. Nothing available this session answers it [UNVERIFIED — no immunogenicity data in patients has been disclosed].

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

Trials

The single most consequential fact in this table, stated plainly and re-verified this sweep. RESTORE-PV’s registered primary outcome measures are all safety — treatment-emergent adverse events by CTCAE grading, abnormal vital signs, abnormal physical examination, abnormal electrocardiograms, and abnormal laboratory results, all over up to 365 days. Every efficacy measure is a secondary outcome, including the one that matters commercially, “proportion of participants achieving therapeutic response, defined as absence of phlebotomy eligibility, during the maintenance period” [VERIFIED — CT.gov get_trial_details NCT06985147, read 2026-08-20]. This is a normal and honest design for a first trial of a long-acting antibody in a chronic disease. But it means that “did the trial hit its primary endpoint?” and “did the drug work?” are two different questions here, and the settlement definition below is written against the second one because that is what the market will price.

Feasibility matrix

Resourcing sufficiency. Not a constraint, and unusually clearly so. The company held $717.7M in cash, cash equivalents and marketable securities at 2026-06-30 against a stated runway “into 2029” and a monthly cash burn of $14.7M (../company.md C.3). Every catalyst in the pipeline is cleared by years on any of the three runway readings, and no registration statement of any kind was filed in 2026. This program will not be starved and the company will not be forced to raise into the readout — which removes the risk that most often converts a good Phase 2 into a flat share price in this corpus. The one thing money cannot buy back is time, and time is exactly what this program is short of relative to its competitors.

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

A.3a Target label. Adults with polycythemia vera who require frequent phlebotomy to maintain haematocrit below 45%, with or without concurrent cytoreductive therapy.

A.3b Target product profile

A.3c Strategic Go/No-Go questions. The pre-Phase-III set is the one that matches this asset’s next decision: RESTORE-PV’s initial data are what the company will use to decide whether to take DISC-3405 into pivotal development in 2027, which is exactly how its chief executive framed it on 2026-07-30 — “we will potentially be positioned to advance two more programs into pivotal-stage development in 2027” [VERIFIED — exhibit 99.1 to the Form 8-K filed 2026-07-30].

A.3d Regulatory designations.

• Orphan Drug Designation (FDA), for DISC-3405 in polycythemia vera, granted 2024-02-09. Grants seven years of US marketing exclusivity for the designated indication from approval, tax credits on qualifying clinical trial costs, and a waiver of the substantial application user fee. It does not speed up review and does not imply any view on efficacy [VERIFIED — company announcement, 2024-02-09, in the BPIQ press feed].
• No other designation. No Breakthrough Therapy, Fast Track, Priority Review or Commissioner’s National Priority Voucher has been announced for this program, and none appeared in the delta window. That is worth noting twice over. This company has obtained a Commissioner’s National Priority Voucher for a different program (bitopertin, October 2025), so the absence here is not for want of familiarity with the route — and the competitor holds Breakthrough Therapy, Fast Track and Orphan Drug designations on the same endpoint in the same disease [VERIFIED — BPIQ press feed and EDGAR filing feed, read 2026-08-20; competitor designations [WEB ESTIMATE — CancerNetwork and OncLive coverage of the rusfertide NDA acceptance, read 2026-08-20]; recorded as an absence rather than as a negative finding].

A.4 Target product profile valuation matrix

Calibration examples, not asset claims: in oncology each incremental month of overall survival is worth roughly $150–300M in peak-sales potential; oral formulations command roughly 15–25% price premiums over injectables. Neither applies directly here — polycythemia vera is a chronic disease managed to a laboratory target rather than a survival endpoint, and every agent in this class is injectable.

A.5 Evidence quality and endpoints

Endpoints

RESTORE-PV’s five primary outcome measures are listed first because they are the trial’s registered primary endpoints, and every efficacy measure below them is a secondary [VERIFIED — CT.gov get_trial_details NCT06985147, read 2026-08-20].

A.5b Key opinion leaders.

Panel as of. 2026-08-20 — the date the investigator and independent-voice searches below were re-run in full.

Investigators

No investigator on this program’s pivotal trial could be named this session either, and that is a finding with a specific cause rather than an omission. get_trial_details on NCT06985147 returns all 15 facilities with contacts: null and carries no overall-official record — the normal state once a trial closes to recruitment, and RESTORE-PV is ACTIVE_NOT_RECRUITING. This was confirmed as a real empty rather than a tool failure by a second, independent call, re-run this sweep with a different result set and the same conclusion: search_investigators on the condition “Polycythemia Vera” across 40 analysed trials returned 72 named investigators and contacts, and not one of them on a Disc Medicine trial. The names returned this time belong to a Phase 3 pelabresib myelofibrosis study (NCT07357727), an ASTX727/iadademstat myeloproliferative-neoplasm study (NCT06661915), a German myelofibrosis research platform and an Egyptian observational study — a completely different set from the 35 returned on 2026-08-13, which is itself informative: the connector is sampling live trials, not failing.

Independent voices

No independent voice is recorded, and the reason is a positive finding rather than a failure to look. The one named panel in the literature that takes a substantive position on this exact endpoint in this exact disease is Marina Kremyanskaya, Yelena Ginzburg and Ronald Hoffman of the Icahn School of Medicine at Mount Sinai, who authored the 2026 Blood review of hepcidin-pathway modulators in polycythemia vera and myelofibrosis (DOI 10.1182/blood.2025028643) and, with Shivani Handa, the 2022 Current Opinion in Hematology review “Hepcidin mimetics in polycythemia vera” (DOI 10.1097/MOH.0000000000000747), whose stated conclusion is that “hepcidin agonists essentially serve as a ‘chemical phlebotomy’ and are poised to vastly improve the quality of life for phlebotomy requiring polycythemia vera patients.”

They are not recorded as independent voices, for two reasons that both point the same way and both survived re-checking this sweep. First, the 2022 review is explicitly a review of rusfertide, describing its global Phase 3 as “currently underway” and advocating for it by name — a published position on a direct competitor to this program, which is a competitor relationship in substance whatever the formal disclosures say. Second, and decisively under rule 40, their conflict-of-interest statements could not be obtained: get_article_metadata returns no conflict field on any of the eight articles retrieved, and both journals are paywalled. Recording them as independent with an empty conflicts array would produce exactly the clean-looking record 02-connectors.md § KOL sources warns against — “the absence of a disclosure is not evidence of no conflict.”

Judgement

Dissent

Empty. Endpoint supported is UNKNOWN rather than SUPPORTIVE_CONTESTED, so there is no majority reading for anyone to dissent from, and inventing a disagreement nobody has expressed would be worse than the blank.

B. Commercial assessment

B.0 Current treatment algorithm

Polycythemia vera is managed to a number. The target is haematocrit below 45%, and everything in the algorithm exists to hold it there.

First, for every patient: low-dose aspirin plus phlebotomy. Aspirin reduces clotting risk. Phlebotomy is the mechanical answer — remove blood, the haematocrit falls. It is cheap, universally available and effective, and it is the reason this disease has been survivable for a century. Its cost is borne by the patient rather than the payer: repeated blood removal strips iron, and iron-deficient patients report fatigue, poor concentration, restless legs and reduced quality of life. Some patients need this every few weeks indefinitely.

Second, for patients at higher risk or needing frequent phlebotomy: cytoreductive therapy. Hydroxyurea is the usual first choice — an oral chemotherapy that suppresses marrow output generally. Ropeginterferon alfa-2b is an alternative, better tolerated by some and worse by others. Ruxolitinib, a JAK inhibitor, is used for patients who fail or cannot tolerate hydroxyurea. All three reduce phlebotomy need by suppressing the marrow, which means they also suppress white cells and platelets, and each carries its own tolerability burden [UNVERIFIED — standard clinical practice, consistent with the treatment landscape described in [DOI 10.1007/s12672-025-03703-9](https://doi.org/10.1007/s12672-025-03703-9), via PubMed].

Where DISC-3405 would fit. In the same slot as rusfertide and divesiran: added to phlebotomy and, where used, to cytoreductive therapy, in order to eliminate the phlebotomies. RESTORE-PV is explicit about this — patients on stable hydroxyurea, interferon or ruxolitinib are eligible, as are patients on phlebotomy alone [VERIFIED — CT.gov eligibility criteria]. It does not displace cytoreductive therapy and it does not treat the underlying mutation. It replaces a procedure, not a drug. That is the commercial proposition and also the commercial constraint: the thing being displaced costs a few hundred dollars.

B.1 Competitive landscape and positioning

Where this asset wins, and the single fact the thesis rests on.

DISC-3405 wins on duration and modality, and on nothing else that is presently demonstrated. An antibody with an eleven-day half-life is a different product from a short-acting peptide injected frequently: fewer injections, steadier exposure, and a dosing schedule that can align with the blood counts these patients already come in for. Against divesiran that advantage has now inverted, and this is the single genuine change in the competitive picture this week. Silence has chosen the every-twelve-week schedule for its Phase 3, not the every-six-week one, despite the six-weekly arm producing the higher response rate (93.8% against 81.3%). Twelve weeks is roughly three times the dosing interval an eleven-day half-life plausibly supports. The convenience argument that was DISC-3405’s only demonstrated differentiator is the argument its nearest competitor has just decided to take to a pivotal trial — and to take there in a randomised, placebo-controlled design this program does not have.

The single fact the thesis rests on is this: the readout must produce a phlebotomy-response number that survives comparison with 76.9% and 88%. Those two figures were produced in placebo-controlled trials against placebo rates of 32.9% and 19%. RESTORE-PV has no placebo arm, so its number will be read by every analyst against those two — and an open-label design is understood to inflate exactly this kind of endpoint, because the decision to perform a phlebotomy is made by an unblinded clinician who knows the patient is on the drug. A 70% response rate from RESTORE-PV is a genuinely good result that will nonetheless be reported as “below rusfertide.”

One competitive fact cuts the other way and is worth stating in its own sentence. Two successful placebo-controlled trials in this indication in twelve months have converted the hepcidin pathway from a hypothesis into an established mechanism. A third drug against a validated mechanism has a much higher probability of technical success than a first-in-class one — which is why the outcome probability below sits at 78% while the commercial assessment stays cautious. This program is likely to work and unlikely to be first. Those are not in tension; they are the whole shape of it.

Calibration example, not an asset claim: oncology first-movers in novel mechanisms have shown roughly 3.2× higher peak-sales potential but about 1.8× higher development risk. DISC-3405 sits on the opposite side of that trade — lower development risk, lower peak-sales potential — which is the correct reading of a third entrant against a validated target.

B.2 Addressable market

Launch markets. The United States first, then EU5 and Japan — the territories the Mabwell licence covers. Greater China and Southeast Asia are explicitly excluded from Disc’s rights and are retained by Mabwell, which is running its own anti-TMPRSS6 antibody, 9MW3011, in polycythemia vera in China [VERIFIED — FY2025 10-K; CT.gov NCT06752746].

Quantified where it can be, and stated plainly where it cannot. Third-party forecasts put the seven-major-market polycythemia vera market at roughly $1.9B in 2024 and $2,087M in 2025, rising to $5,257M by 2034, with rusfertide expected to generate the highest revenue among emerging therapies in EU4 and the UK by 2034 and to reach a probability-adjusted peak share of 5.4% in first and second line about seven years after launch [WEB ESTIMATE — DelveInsight polycythemia vera market report, read via web search 2026-08-20; the same figures the previous sweep found, re-checked rather than carried]. These are the only market-size figures found and they are third-party projections, not a bottom-up build; the “5.4% peak share” figure in particular is quoted as it was found and its denominator is not stated in the source.

B.3 Value and feasibility

B.3a Expected peak sales. Built as a share of the class rather than fully bottom-up, because the patient-count input in B.2 could not be established and rule 9 forbids inventing one. The chain is: total polycythemia vera market → the share hepcidin-pathway agents take of it → DISC-3405’s share of that class. Every link is tagged, and the whole build is conditional on approval, which on the current timeline could not occur before roughly 2030. It excludes the sickle cell indication, any ex-US partnering economics, and Greater China and Southeast Asia entirely, which Disc does not own.

The bands are carried unchanged from the previous version and that is a deliberate decision, not inertia. Nothing in the delta window changed the market size, the class share or this asset’s plausible share of it in a way a number could honestly express. What did change is the direction of pressure on the high case, and that is recorded in its Basis cell rather than by moving a figure without evidence.

Which input is weakest, named rather than buried. It is the patient count, and everything downstream inherits its uncertainty. No defensible epidemiology figure for the phlebotomy-dependent polycythemia vera population in the launch markets was obtained this session, so the build routes around it through a third-party market projection whose own methodology is not published. The share assumptions are analyst judgement with no external anchor at all. These are ranges to reason with, not figures to value on, and no single point estimate is given anywhere (rule 10).

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Partly, and less than it did a week ago. RESTORE-PV is well designed to answer whether the drug controls haematocrit and eliminates phlebotomies in patients, and to select a dose. It is not designed to support the differentiation claim the commercial case rests on. The target product profile’s distinguishing feature is dosing convenience versus a short-acting peptide, and an open-label single-arm trial with no comparator cannot demonstrate an advantage over rusfertide on anything. That comparison will have to be made across trials, which payers and prescribers discount heavily — and the competitor now going to Phase 3 will be able to make its own convenience claim inside a randomised design.
2. Will the identified risks affect the target product profile? The two that would are the therapeutic window and immunogenicity. If iron restriction has to be dialled back to avoid symptomatic deficiency, the response rate falls and the profile converges on “another option”. If anti-drug antibodies blunt the effect over time, the long-half-life advantage inverts into a durability problem. Neither is answerable from this readout alone.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? On modality, yes — it would remain the only antibody in the ex-China class. On anything a prescriber or payer chooses on, no. Against divesiran’s twelve-weekly siRNA, an antibody’s duration advantage is not merely thin, it is the wrong way round, and the fallback differentiation is a preference for a familiar modality rather than a clinical claim.

Readout

Sources

The modelled estimate. The usual arithmetic — registry primary completion plus the default lag — cannot be applied to this catalyst, and saying so is more useful than forcing it. RESTORE-PV’s primary_completion_date is 2027-02; primary completion plus two to four months gives 2027-04 to 2027-06, which is seven to nine months after the guided event. That is not a disagreement with the company, it is the default lag being applied to the wrong endpoint: rule 34’s arithmetic estimates a final readout from a final completion date, and this catalyst is an interim cut.

The estimate is therefore built from the operational fact the company disclosed instead. Enrolment completed on or before 2026-07-30 (announced that day). Last patient in plus the default two-to-four month lag to data cleaning, database snapshot and analysis gives 2026-09-30 to 2026-11-30, narrowed at the front to 2026-09 because early-enrolled patients have been dosed since August 2025 and an interim cut does not wait for the last patient to mature. Tagged [UNVERIFIED — modelled, default lag] rather than benchmarked because data/benchmarks/readout-lag.json still holds zero observations — nothing in this repository has been confirmed to benchmark against, re-checked this sweep. Note the direction: the modelled range’s tail runs two months past the guided quarter, which is what pushes the window’s latest edge out below.

Window

Basis. Precision is PERIOD because the only source that discloses anything names a quarter — “Q3 2026” — and no source names a month or a day; that is the mechanical test in rule 23 and it is what holds the run-up priority score down to 12 below. The likeliest date sits in the last week of September because the guided quarter ends 2026-09-30 and a company that says “ahead of schedule” while still guiding to a quarter most often lands inside its final month. The latest edge runs one month past the guided quarter, taking the front half of the modelled range’s tail, because an interim cut can slip without the company having missed anything it promised.

The earliest edge is re-derived rather than carried, and it has got stronger rather than weaker. The 2026-08-13 version flagged 2026-09-01 as the weakest judgement in the document, because it assumed nothing would print in the remaining two and a half weeks of August — an assumption about disclosure habits rather than a fact. Seven days later that assumption covers a much smaller and much better-defended interval: seven trading sessions, 2026-08-21 to 2026-08-31, which are the thinnest disclosure window in the US biotech calendar. The company reaffirmed Q3 on 2026-07-30 without naming a nearer date and has said nothing since. So 2026-09-01 stands, and it stands on a narrower claim than it did. A late-August print remains possible and would resolve the run-up exit rule below to a date already in the past — that risk has not vanished, it has shrunk, and it is stated here rather than hidden inside the entry.

Disagreement. CONSISTENT. The three sources that speak to this catalyst do not conflict; they answer different questions. BPIQ’s 2026-09-30 is the synthesized end of the quarter the company named, so it agrees with the company by construction. The CT.gov 2027-02 is the trial’s final primary completion and is not a competing estimate of this event at all. The modelled 2026-09 to 2026-11 overlaps the guided quarter and extends past it. No source points earlier than any other in a way that needs resolving. One near-miss is recorded rather than dressed up as a conflict: a third-party aggregator page surfaced during this sweep’s web searches described DISC-3405’s initial data as expected in “Q4 2026”. It is not a source — it carries no date of its own, no attribution and no company quotation, and the same page when fetched directly gives no catalyst date at all. It is stale copy predating the 2026-07-30 pull-forward, and it is named here so that a later reader who meets it does not mistake it for a disagreement this analysis failed to notice.

Date slippage. Zero slips, and one move in the opposite direction. The BPIQ note field carries a single dated entry for this row, re-read this sweep and byte-identical to the previous one, and the company’s own releases carry no earlier guidance that was abandoned. The one recorded change is a pull-forward, not a slip.

Zero slips is a finding, and on this ticker it is a distinguishing one: bitopertin’s FDA decision has moved by more than a year over the same period, and the same 2026-07-30 release pushes it again, to “mid-2027”.

Attribution

Status.

Conflicts

Computed by lib/clustering.mjs’s attributionFor('IRON', companyRecord, programs, 6, 18007), re-run over this ticker’s full nine-row pipeline table on 2026-08-20 and transcribed rather than judged by eye. The result is identical to the previous sweep, because neither the pipeline rows nor this program’s own window moved.

Three of the four other has_catalyst rows on this ticker collide with this program’s window; the fourth (19456, DISC-0974 in inflammatory bowel disease, guided to 2027) sits outside the six-month bar. Every gap is zero days — the ranges overlap outright — and not one conflict is confirmed, because this program’s own window carries PERIOD precision and all three colliding rows carry catalyst_date_is_exact: false. Two guesses touching is evidence of not knowing.

Note. Not required for INDETERMINATE, and none is given. The stock-direction call below states the consequence in prose instead, as rule 36 requires.

A reading the computation cannot express, recorded because it matters more than the status does. INDETERMINATE is the correct computed answer and it understates the real situation. The colliding rows are unconfirmed only because every date on this ticker is a period-end placeholder — but the largest of them, bitopertin’s APOLLO topline, is guided to Q4 2026, and the options market has already priced it: at-the-money implied volatility roughly doubles between the October and November expiries (../company.md C.6). Any position held past early October is exposed to an event that the market itself says is roughly twice as large as this one. The stock-call window below is closed on 2026-10-14 for exactly that reason, and that closure is the honest response to a collision the clustering computation is structurally unable to confirm.

Market and timing for this event

• Plain takeaway. The market paid for part of this three weeks ago and has been marking it down since. The shares rose 7.5% on 2026-07-30, the session the readout was pulled forward into Q3, and are up 7.9% from $73.67 (2026-07-29) to $79.51 today — but the options market’s own expected move into mid-September has fallen from 12.0% to 10.4% in seven days, at-the-money open interest on the September expiry has not gained a single contract, and average dollar volume is down about a fifth. What is left to price is the result itself, into a chain that says this is the smaller of the two events on this ticker and is saying it slightly more quietly than last week.
• Months to this catalyst. From readout.window.earliest of 2026-09-01: 0.4 months, twelve calendar days, seven trading sessions. readout.precision is PERIOD, so that figure is derived from the start of a judged window inside a guided quarter, not from a disclosed date — and per the Readout basis above, a print inside those seven sessions would make it zero.
• Expected move around this event. ../company.md C.6 records the chain as readable but thin. The 2026-09-18 at-the-money straddle divided by spot is 10.4%, down from 12.0% on 2026-08-13, with a liquidity confidence of low-to-medium and a bid-ask spread of roughly 20% of the midpoint on the $80 call. Read it as a bracket of roughly 9–12%, not a point estimate, and note that it covers everything between now and 2026-09-18, not the readout alone. For context rather than as a substitute, the published class average for a Phase 2 readout by an emerging biopharma company is +12% on a positive result and −16% on a negative one [WEB ESTIMATE — IQVIA 2024] (framework/07-benchmarks.md). The chain and the class average now disagree on magnitude as well as on symmetry: the chain prices ±10.4%, the class average is asymmetric at +12%/−16%, and the chain is symmetric by construction.
• Nearest comparable past reaction. 2026-06-12 is the closest structural analogue — a hepcidin-pathway data readout on this ticker — and it produced −0.13% (../company.md C.7). But the right analogue is 2024-06-14, the MWTX-003 Phase 1 polycythemia vera data at EHA: +18.16% close to close on six times normal volume, achieved on a session that also absorbed a discounted equity offering. It is the right one because it is the same molecule in the same indication, and it is the only reading in ten years of catalyst history where a non-bitopertin program moved these shares more than 10%. It is also two years old, from a much smaller company at a $1.4B rather than a $3.1B market capitalisation, and from a Phase 1 in healthy volunteers rather than a Phase 2 in patients — so it is the ceiling the positive range stops short of, not its centre. That difference in company size is a calibrated point rather than a rhetorical one: framework/07-benchmarks.md puts Phase 2 reactions at emerging biopharma companies under $1B market capitalisation at 9.5 times those at or above it [WEB ESTIMATE — IQVIA 2024], and this ticker crossed that line between the two events.
• Materiality. Meaningful, not dominant, as recorded in ../company.md C.2. The stock-direction call below is consistent with that: a modest directional call inside a narrow band, not a doubling or a halving.
• Date slippage. Zero slips, per Readout above — and the one recorded change is a pull-forward. Zero slips is a finding, and here it is a favourable one.

Spot. $79.51, the value carried in data/prices/IRON.json for 2026-08-20, read through lib/prices.mjs [VERIFIED — data/prices/IRON.json via lib/prices.mjs, read 2026-08-20]. It is a partial intraday bar, not a close, written while the session was open on 106,841 shares against a typical full-day volume of about 335,000; BPIQ read $78.92 the same day, 0.7% lower, also intraday (../company.md C.1 and C.8). Every percentage below is measured against $79.51 because that is what a settlement replayed against the committed cache will see, and the 0.7% uncertainty is disclosed rather than buried.

One correction to the superseded chain, because it changes a figure that was carried forward. The 2026-08-13 record locked at “$79.50, the closing price on 2026-08-13”. That figure was also an intraday snapshot: the cache stored it on 66,796 shares, and today’s fetch reported an $80.01 close on 341,000 shares for the same date and discarded it, because scripts/fetch-prices.mjs never overwrites a non-null value. The 2026-08-13 lock was therefore struck 0.6% below that day’s real close, and every percentage anchor in it inherited the error. The 2026-08-15 re-lock’s spot of $78.84 (the 2026-08-14 close) is not affected — that bar is complete, on 262,500 shares.

How these two ranges are calibrated. The anchor set is the same four per row the previous version used, but three of the four have moved and are re-read rather than re-based: the options-implied move has fallen, the named analyst target has risen, and spot itself has changed. The shape is re-derived under framework/07-benchmarks.md, whose three relevant readings do not all point the same way:

1. Phase 2 asymmetry. Negative readouts move emerging-biopharma shares more than positive ones at every phase; at Phase 2 the class averages are +12% and −16%, a ratio of 1.3 [WEB ESTIMATE — IQVIA 2024]. The file states plainly that a miss range set no deeper than the positive range is high needs a program-specific reason, and that an expected value computed off symmetric ranges is biased optimistic by exactly that asymmetry.
2. Therapy area. Polycythemia vera is a myeloproliferative neoplasm, so the applicable row is oncology — the only near-symmetric therapy area in the table, +9% against −9%, ratio 1.0 [WEB ESTIMATE — IQVIA 2024]. The file attributes that near-symmetry to the high share of single-arm trials in oncology, which is exactly what RESTORE-PV is.
3. Trial design. The design being priced is this program’s own pivotal-intent trial, and it is single-arm, whose class impact amplitude is 7% — against 29% for a placebo-controlled trial and 25% across the whole trial universe [WEB ESTIMATE — IQVIA 2024]. The file is explicit that the single-arm rows did not reach statistical significance, so this is a direction, not a number: it argues for compression in both directions, not for collapsing the bands to 7%.

Where a specific fact beats the class average, the specific fact wins, which is the file’s own rule. Two do so here, and one of them changed. The readable options chain now prices a ±10.4% move rather than ±12%, which pulls both rows in. And this ticker’s own C.7 history keeps the miss floor off the class asymmetry: the deepest catalyst reaction on this ticker (−21.91%) belongs to bitopertin, the dominant program, and this program’s materiality is “meaningful, not dominant”. The net effect against the superseded record is that the positive top comes in (from +16.7% to +14.5%) and the miss floor comes up (from −17.6% to −15.7%), both driven by the same fact — the market is pricing a smaller September move than it was. The asymmetry runs at 1.08, essentially where the re-lock put it, and still short of the all-therapy-area Phase 2 figure of 1.3 for the oncology and single-arm reasons already recorded. Nothing in framework/07-benchmarks.md is used as an anchor; rule 30’s two named anchors per range are satisfied four times over in each row.

Scenario prices

Expected value. $83.76, +5.3% against spot of $79.51. Method: the 78% probability applied to the midpoint of the positive range ($87.50) and the residual 22% to the midpoint of the miss range ($70.50) — 0.78 × $87.50 + 0.22 × $70.50. 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.

The superseded record put this at $83.93, +6.5% against a spot of $78.84. The dollar figure is almost unchanged and the edge has fallen by 1.2 points, and the two facts have the same single cause: spot rose 0.9% while the option market’s priced move fell 1.6 points, so both scenario rows moved toward spot and the gap between the probability-weighted midpoint and the price narrowed. The call stays up because +5.3% still agrees with the direction, but it is a smaller edge than the record it replaces, for a reason that is readable off a quoted straddle rather than off a judgement.

Run-up

Entry and exit

This is no longer really a run-up leg, and the honest thing is to say so in the same breath as locking it. readout.window.earliest is 2026-09-01, so the exit rule resolves to approximately 2026-08-25 — three trading sessions after entry (21, 24 and 25 August), against seven in the 2026-08-15 re-lock and eight in the original 2026-08-13 lock. The leg has shortened on every successive lock for the same mechanical reason: the window is fixed and the lock date keeps moving toward it. Three consequences follow and all three are recorded rather than smoothed over.

First, most of the run-up has already happened without this position — the shares gained 7.5% on 2026-07-30 when the readout was pulled forward and are 7.9% above their pre-announcement level, and the option market has been marking the coming move down rather than up for a week.

Second, a print inside the seven remaining August sessions would resolve this exit rule to a date already past, which is the concrete cost of a PERIOD-precision window and is exactly what the date_confidence driver below is measuring.

Third, and most simply, three sessions is not enough time for a pre-event drift to express itself. The call is locked because rule 39 withholds a run-up only when date_confidence floors at zero — it scores 20 here, not 0 — but a reader should treat the band below as a near-flat placeholder rather than as a trade thesis. The scored call this document actually stands behind is the outcome call.

exit itself is null at lock time. lib/runup.mjs’s resolveExit returns null for all three rules against the committed price cache, because the cache ends on 2026-08-20 and there are no bars at or after readout.window.earliest to count back from. That is the normal state for a run-up call at lock and is recorded rather than estimated.

Predicted move and peak

Priority score drivers

Priority score. Priority score 12 · formula_version 1.0.0

Computed by lib/runup.mjs’s priorityScore over the seven drivers above, never hand-computed, and re-run in full for this record at the unchanged formula_version 1.0.0. It returns 12, against 11 for the superseded record. The whole of the one-point move is the squeeze driver’s volume-bucket flip, partly offset by priced_in falling a point: the unrounded score is 11.51 against 11.35, so the change is 0.16 of a point that rounding happens to expose, not a re-rating. For context on what a 12 means: date_confidence of 20 both multiplies the weighted base by 0.20 and triggers the formula’s untradeable cap of 15, so a program with an unusually strong squeeze setup and a 78% outcome probability still ranks low. That is the intended behaviour, not a defect — the score is saying that a trade you cannot time is not a good trade however good the asset is. It is worth being exact about which of the two mechanisms binds: at 11.51 the untradeable ceiling of 15 is not binding; the 0.20 multiplier is. Note also that framework/07-benchmarks.md is not read by this formula and no driver here was changed by it — folding a class prior into the seven drivers would be a formula_version bump, which is a separate decision that has not been taken.

Settlement. Left null at lock time. Settled only on an explicit user request, from a confirmed primary source, against the committed price cache.

Verdict

What I would do. Watch, and treat the run-up leg as effectively closed rather than as a second call worth acting on.

Why. The science here is in better shape than almost anything else in this corpus: the mechanism is fully mapped, every step of it was measured in humans in a peer-reviewed Phase 1, the target has a Mendelian human phenotype, and two independent competitors have now proved the pathway works in this exact disease in placebo-controlled trials. That is why the outcome probability is 78% rather than a coin flip. The problem is not whether the drug works — it is that the same competitor success that de-risks the mechanism also caps what the result is worth, and this week it capped it a little further. Rusfertide holds an FDA decision this quarter, still pending; divesiran posted 88% versus 19% and has now chosen an every-twelve-week schedule for its Phase 3, which takes the dosing-convenience argument that was DISC-3405’s only demonstrated differentiator and puts it inside a randomised design at an interval an eleven-day half-life cannot match. An open-label, single-arm, 60-patient result whose registered primary endpoint is safety will be read against those two numbers. Layered on top, this is the second program on a ticker whose equity trades on bitopertin, whose Phase 3 lands one quarter later and which the options chain prices at roughly double the implied volatility. Attribution could not be determined for this program — three other catalysts on this ticker overlap this window at zero days’ gap, and none of the collisions can be confirmed because every date Disc publishes is a quarter or a year rather than a day.

What would change this. Two observables, in opposite directions. To the upside: a phlebotomy-response rate at or above 77% in the maintenance period with a clean safety profile and a dosing interval of four weeks or longer. That combination would put DISC-3405 level with rusfertide’s active arm on efficacy while beating it on burden, which is the only version of this story that supports the high peak-sales case rather than the base one. To the downside: an initial cut that reports only pharmacodynamics — hepcidin up, iron down, transferrin saturation down — with no phlebotomy or haematocrit outcome in patients. That is not a miss, but it is a non-event dressed as a catalyst, and it would leave the differentiation question open for another year while two competitors move.

What to watch.

• From 2026-08-21, any day. The readout itself. Read the phlebotomy-response rate first, and read which period it covers. “Absence of phlebotomy eligibility during the maintenance period” is the endpoint with the external benchmark; the optimization-period version of the same endpoint is lower by construction because patients are still being titrated, and an initial cut has every reason to report the one that is ready rather than the one that matters.
• On the same day. The evaluable n. Sixty patients enrolled does not mean sixty patients analysed in an interim cut, and a response rate on twelve patients is a different fact from the same rate on forty.
• On the same day. The dosing interval actually used. It is the entire differentiation argument against rusfertide, it has never been disclosed, and divesiran’s Phase 3 has just set the bar at twelve weeks.
• On the same day. Any safety finding involving symptomatic iron deficiency, or any cohort-wide dose reduction. With an eleven-day half-life this is the near-veto risk, and it is the trial’s registered primary endpoint.
• Q3 2026, date not disclosed, now inside weeks. Rusfertide’s FDA decision. An approval sets the price, the label and the reimbursement precedent DISC-3405 will inherit; a Complete Response Letter would reset the regulatory bar for the whole class and would matter more to this program than its own readout does.
• Around 2026-08-26. The FINRA short-interest settlement for 2026-08-14, which will be the first observation of whether the shorts kept adding through the first half of August. Short interest is one of the two computed drivers behind this program’s only genuine tailwind (../company.md C.5).
• Q4 2026, date not disclosed. Bitopertin’s APOLLO topline, with a Complete Response Letter response and an FDA decision now guided to mid-2027. Not this program’s event, but the one that decides this equity — and the reason the stock-call window below closes on 2026-10-14 rather than running to the end of the readout window.
• An upcoming medical congress, date not announced. Silence Therapeutics said it will present full SANRECO results for divesiran. The topline is already the benchmark; the full dataset will set the bar DISC-3405’s own full dataset has to clear.
• Any date before the readout. An 8-K or 424B5 indicating a financing. None is expected — $717.7M, runway into 2029, no registration statement filed in 2026 — so one would be genuinely informative rather than routine.
• Any date. A first open-market purchase by any Disc insider. Thirty-six Form 4s in 2026 and not one buy (../company.md C.5); a buy would be the first insider signal in the record that is not a scheduled sale. Note that the chief executive’s $2.61M sale on 2026-08-17 is not a signal in the other direction: it ran under a Rule 10b5-1 plan adopted 2026-03-12, five months before the trade and before the readout was pulled forward.

Locked prediction

• What this record is. IRON-18007-2026-08-20, a full refresh, superseding IRON-18007-2026-08-15 (itself a re-lock of IRON-18007-2026-08-13). Both tiers of the sweep were re-run in full on 2026-08-20 and a new coverage record was earned, which is what distinguishes this from the record it replaces.
• Outcome-direction (will the readout succeed on its primary endpoint?): positive — probability 78%, band 65–87% [UNVERIFIED — modelled]. The figure is carried unchanged from the superseded record, and that is a finding rather than an omission: a full re-sweep of both tiers found nothing that moves it. The registry did not change, the literature did not grow, the company said nothing, and the two competitor developments in the delta window (rusfertide’s decision still pending, divesiran’s Phase 3 scheduled) bear on what a positive result is worth, not on how likely it is. The reasoning, tied to the settlement definition below and to nothing vaguer: the drug’s full pharmacological chain is already demonstrated in humans and published; the target is validated by a Mendelian human phenotype, by a JAK2-mutant mouse model, and by a second drug against the identical target meeting its Phase 2 primary endpoint in this indication; the trial is open-label and single-arm against each patient’s own documented phlebotomy history, a design that flatters this endpoint rather than challenging it; and the company pulled the readout forward a quarter, which sponsors do far more often when emerging data are good than when they are not. The 22% that remains is roughly two parts an initial cut too immature to report a maintenance-period response, and one part a safety finding — the drug’s therapeutic effect and its principal expected toxicity are the same physiological change, and an eleven-day half-life makes it slow to reverse. No third-party published probability on this specific event was found this sweep either.
• Stock-direction (which way do the shares move?): up — confidence low — window 2026-08-20 to 2026-10-14, basis: from this record’s own lock date through two weeks past the end of the guided quarter, deliberately stopping short of readout.window.latest of 2026-10-31 because bitopertin’s APOLLO topline is guided to Q4 2026 and the options chain prices it at roughly double this event’s implied volatility, so any move after early October would not be attributable here. The window now opens on the lock date rather than on 2026-08-13, because a refresh re-opens the call where a re-lock does not. Attribution could not be determined for this program (rule 36): three other has_catalyst rows overlap this window at zero days’ gap and not one collision is confirmable, because every catalyst date on this ticker is a period-end placeholder rather than a disclosed day. Materiality is meaningful, not dominant (../company.md C.2), and the call is sized to match — a modest directional view, not a re-rating.
• Scenario prices: positive $84–$91 · miss $67–$74 (re-derived on a spot of $79.51 and a 10.4% options-implied move; was positive $84–$92, miss $65–$74 on a spot of $78.84 and a 12.0% move)
• Expected value: $83.76, +5.3% against spot $79.51 (arithmetic, not advice; was $83.93, +6.5%)
• Run-up: entry $79.51 on 2026-08-20, exit rule T-5 trading days before readout.window.earliest — predicted move 0–4%, predicted peak 2–14% around 2026-09 — priority score 12, formula_version 1.0.0. The leg is three trading sessions long and the Run-up section above says plainly that it should not be traded on.
• Settles on the first public disclosure of initial data from RESTORE-PV (NCT06985147), by Disc Medicine press release, Form 8-K, or Q3 2026 quarterly results, whichever comes first. Positive is defined as both of the following. (a) An efficacy signal in evaluable polycythemia vera patients: either a therapeutic-response rate (absence of phlebotomy eligibility) of 60% or more — a threshold set against the placebo rates actually measured in the two controlled trials of this class, 32.9% in VERIFY and 19% in SANRECO, so 60% is roughly double the higher of them — or a reported reduction in phlebotomy rate the company characterises as clinically meaningful, or a majority of evaluable patients maintaining haematocrit below 45% without phlebotomy. And (b) no new safety finding leading to a clinical hold, a cohort-wide dose reduction, suspension of dosing or discontinuation of the polycythemia vera program, with the company stating it intends to advance DISC-3405 in polycythemia vera. A disclosure reporting only pharmacodynamic changes (hepcidin, serum iron, transferrin saturation) with no phlebotomy or haematocrit outcome in patients does not meet this definition. (Carried verbatim from the superseded chain on purpose: a settlement definition that drifts between locks destroys the comparability the chain exists to provide, and nothing this sweep found argues the definition was wrong.)
• Locked: yes · Settled: no

Program data-quality flags

• The entry price is a partial intraday bar, and so was the 2026-08-13 lock’s. scripts/fetch-prices.mjs never overwrites a non-null value, so a bar first written mid-session stays an intraday snapshot for ever. Today’s $79.51 was written on 106,841 shares against a typical 335,000, and BPIQ read $78.92 the same day. The consequence for the superseded chain is specific: the 2026-08-13 record’s “$79.50 close” was itself an intraday snapshot on 66,796 shares, against a discarded $80.01 close on 341,000 shares, so that lock was struck 0.6% below the real close and every percentage anchor in it inherited the error. The 2026-08-15 re-lock’s $78.84 is unaffected — that bar is complete. Full detail in ../company.md C.8.
• Open Targets’ search layer answers and its GraphQL layer does not. search_entities resolved TMPRSS6 to ENSG00000187045, polycythemia vera to MONDO_0009891 and HAMP to ENSG00000105697, so the mandatory row is CALLED — an improvement on the previous sweep, which recorded it BLOCKED. But three attempts at the association query returned verbatim Rate limit exceeded for client: global, so there is still no independent human-genetics association evidence for TMPRSS6 in this document. Resolving an identifier is not validating a target, and the A.1 claim stays [UNVERIFIED].
• The registry’s primary_completion_date does not date this catalyst, and using it would be wrong by seven to nine months. RESTORE-PV’s 2027-02 is the final primary-completion date for a 365-day safety endpoint; the guided event is an interim cut. Rule 34’s default-lag arithmetic applied to it gives 2027-04 to 2027-06. The modelled estimate is built from the disclosed enrolment-completion date instead, and the substitution is stated in the Readout section rather than performed silently.
• RESTORE-PV’s registered primary endpoints are all safety; every efficacy measure is secondary. “Did the trial hit its primary endpoint?” and “did the drug work?” are therefore different questions on this program. The settlement definition is written against the second because that is what the market will price, and the divergence is flagged here so a later reader does not mistake it for an error.
• No efficacy data exists for DISC-3405 in any patient with any disease. Every efficacy figure in this document belongs to a competitor. The Phase 1 was in healthy volunteers.
• The only publication on this molecule has no independent author. All five authors of the Phase 1 paper give their affiliation as “Disc Medicine, Watertown, MA, USA”, re-confirmed from the metadata this sweep. The independent literature on hepcidin-pathway agents in polycythemia vera — specifically the 2026 Blood review — names rusfertide, divesiran and sapablursen and does not name DISC-3405, while naming Disc’s other antibody in its myelofibrosis section; that sentence was read directly from the abstract rather than inferred.
• No investigator on RESTORE-PV could be named, and no independent voice could be recorded. Both empty arrays carry their searches on record in A.5b, re-run in full this sweep with a different 72-investigator result set and the same conclusion. kol.judgement.endpoint_supported is consequently UNKNOWN.
• The patient-count input to B.3a could not be established and no figure is invented. The peak-sales build routes around it through a third-party market projection whose methodology is not published, and the share assumptions inside it have no external anchor at all. This is the weakest input in the document and B.3a says so in its own text.
• 9MW3011 may or may not be the same molecule as DISC-3405. Mabwell retains Greater China and Southeast Asia under the licence and is running an anti-TMPRSS6 antibody in polycythemia vera in China (NCT06752746). Whether it is MWTX-003 under a Chinese development code, or a different antibody from the same family (MWTX-001 or MWTX-002 are also licensed), could not be established this session either and remains an open question rather than something resolved by assumption.
• A stale third-party page describes this readout as “Q4 2026”. It surfaced in this sweep’s web searches, carries no attribution, no date of its own and no company quotation, and gives no catalyst date at all when fetched directly. It is copy predating the 2026-07-30 pull-forward. It is recorded here, and in the Readout disagreement note, so that a later reader who meets it does not mistake it for a conflict this analysis failed to see.
• Analyst price targets moved inside the delta window and they are company-level, not program-level. Morgan Stanley raised from $85 to $90 on 2026-08-17 and H.C. Wainwright from $118 to $120 on 2026-08-10; the thirteen-analyst consensus is $101.17 with a $128 high and a $79 low, essentially spot [WEB ESTIMATE — stockanalysis.com, last updated 2026-08-17, read 2026-08-20]. Every one of these targets is dominated by bitopertin’s APOLLO readout rather than by this program, which is why the positive scenario row uses them as bounding anchors and not as a valuation.
• The company’s cash-runway guidance, BPIQ’s, and the arithmetic disagree (“into 2029”, 2030-06-30, and 2029-07 to 2030-07 respectively). Reported as a disagreement in ../company.md C.3 rather than averaged; the conservative reading is used throughout. It does not bear on this program’s conclusions because every reading clears this catalyst by years.
• The readout.window.earliest of 2026-09-01 is still the weakest judgement in this document, but it is a smaller claim than it was. It now asserts only that nothing prints in the seven remaining August trading sessions, the thinnest disclosure window in the US biotech calendar, rather than in two and a half weeks including the more active first half of the month. A late-August print would still resolve the run-up exit rule to a date already past. Stated in the Readout basis and again in the Run-up section rather than left for a reader to discover at settlement.