EVMN / evo756-atopic-dermatitis — EVO756 for moderate-to-severe atopic dermatitis (eczema)

Program analysis · bpiq_drug_id 20071 · prepared 2026-08 · USD · framework v5.8.0 · 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].

Glossary

Executive summary

• What it is (one sentence): EVO756 is a pill that blocks MRGPRX2, a switch on mast cells that makes them release histamine and other irritants without any allergic antibody being involved, and it is being tested to see whether shutting that switch off reduces eczema.
• The event and when (as disclosed): Top-line data from a fully enrolled Phase 2b dose-ranging trial in 120 adults with moderate-to-severe atopic dermatitis. The company said on 2026-08-06 that it expects to report in September 2026 [VERIFIED — Evommune Q2 2026 release]. BPIQ carries the vaguer “Q3 2026”. Neither names a day.
• The main reason it could work: MRGPRX2 is a genuinely well-validated driver of non-allergic mast-cell activation and of itch that antihistamines do not touch, the target is supported by independent structural and genetic work, and blocking it improved eczema-like disease in animal models. EVO756 has shown it engages the target in humans.
• The main risk, and it is the whole analysis: the same molecule already failed the same kind of trial, at every dose, in the disease where this mechanism is strongest. On 2026-06-29 the Phase 2b in chronic spontaneous urticaria missed its primary endpoint at all three doses and the indication was discontinued. Urticaria is the flagship mast-cell disease. Atopic dermatitis is a barrier-and-type-2-inflammation disease in which mast cells are a contributor rather than the engine. Succeeding here after failing there requires the mechanism to matter more in the disease where it plausibly matters less.
• What it means for the stock: The shares have already fallen 57% from their 52-week high and trade at 1.36× cash per share, so a lot of bad news is in the price. The probability-weighted arithmetic lands within 2% of spot. No edge on direction — but a wide two-sided outcome, and a company that survives either result.

0. Program-tier coverage — CLEARED

Company-tier coverage is in ../company.md C.0.

What the ChEMBL empty costs. ChEMBL would have given EVO756’s structure and its off-target binding profile, which is the usual way to ask “does this molecule hit anything besides MRGPRX2?”. Without it, every selectivity claim in this document rests on the company’s own description of EVO756 as “highly selective” [UNVERIFIED — company website, not independently confirmed]. The practical cost is low, because the safety record from three completed human trials is the better evidence on that question and it is clean; the cost is real for the question of whether the CSU failure could reflect insufficient target coverage rather than a wrong target, which no source this sweep reached can answer.

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

What the drug is. EVO756 is a small molecule taken by mouth. It blocks a receptor called MRGPRX2 [VERIFIED — Evommune corporate materials; ClinicalTrials.gov NCT07150845].

How it works, plainly. Mast cells are immune cells stationed throughout the skin, sitting right next to the endings of sensory nerves. They are loaded with granules of histamine, tryptase and other irritants. Everyone knows one way to set them off: an allergen binds to IgE antibodies on the cell surface and the cell empties its granules. MRGPRX2 is a second, separate switch, on the same cells, that does the same thing without any antibody involved. It is triggered by a broad set of positively charged molecules, including substance P and other peptides that nerves release when skin is inflamed, and antimicrobial peptides that skin cells themselves make [VERIFIED — Roy et al., J Allergy Clin Immunol 2021, [DOI](https://doi.org/10.1016/j.jaci.2021.03.049); Cao et al., Nature 2021, [DOI](https://doi.org/10.1038/s41586-021-04126-6). According to PubMed.].

That gives a plausible vicious circle in eczema: inflamed skin makes nerves release peptides → peptides switch on MRGPRX2 → mast cells degranulate → itch → scratching → more barrier damage and more inflammation. EVO756 is intended to break the circle at the MRGPRX2 step.

How well the target is validated — and what kind of validation it is. Strongly, but almost entirely in the laboratory and in animals, and almost none of it in atopic dermatitis in people.

• MRGPRX2’s structure has been solved and potent antagonists designed against it by an independent academic group [VERIFIED — Cao et al., Nature 2021, [DOI](https://doi.org/10.1038/s41586-021-04126-6). According to PubMed.]
• The mouse equivalent of the receptor drives itch that is genuinely independent of histamine and IgE, and deleting it reduces itch in models of allergic contact dermatitis [VERIFIED — Meixiong et al., Immunity 2019, [DOI](https://doi.org/10.1016/j.immuni.2019.03.013). According to PubMed.]
• A different small-molecule MRGPRX2 antagonist, GE1111, improved eczema-like skin disease in mice, reduced the relevant inflammatory signals and preserved skin-barrier proteins [VERIFIED — Wong et al., Front Immunol 2024, [DOI](https://doi.org/10.3389/fimmu.2024.1406438). According to PubMed.] This is the single closest published support for this exact program, and it is a mouse study run by a rival academic group on a rival compound.
• The competitor’s own antagonists blocked mast-cell degranulation in human skin tested outside the body [VERIFIED — Wollam et al., J Allergy Clin Immunol 2024, [DOI](https://doi.org/10.1016/j.jaci.2024.07.002). According to PubMed. Note: every senior author is an Escient Pharmaceuticals employee — the competitor.]
• Open Targets confirms MRGPRX2 as a recognised human target, ENSG00000183695 [VERIFIED — Open Targets search_entities 2026-08-13].

The exact scientific step the next readout must prove. That blocking this one non-IgE mast-cell route produces a statistically significant reduction in the percentage change in EASI at week 12 in moderate-to-severe atopic dermatitis. Not that MRGPRX2 exists, not that EVO756 blocks it, not that mast cells matter in eczema — all three are established. The open question is whether this lever is big enough to move this disease.

The honest scientific risk, stated in full. Two things point the wrong way and they compound.

First, the disease biology is a poor fit for the mechanism’s strongest case. Atopic dermatitis is driven by a defective skin barrier and by type-2 inflammation — interleukin-4, interleukin-13 and interleukin-31. Every approved biologic for the disease targets that axis, and does so successfully. Mast cells contribute to the itch and to the inflammatory tone, but they are not the engine. Chronic spontaneous urticaria is the opposite: it is a mast-cell disease more or less by definition, which is why MRGPRX2 programs are aimed there first.

Second, and decisively, the molecule has already been tested against that stronger case and lost. The Phase 2b in chronic spontaneous urticaria enrolled 182 patients across 53 sites in four territories, ran three dose regimens against placebo for twelve weeks, and failed to separate from placebo on UAS7 at any dose [VERIFIED — Evommune press release 2026-06-29; ClinicalTrials.gov NCT06873516]. The company’s own Chief Medical Officer said the lack of efficacy “does not support further development of EVO756 for CSU,” while noting that the trial had shown safe, tolerated doses and demonstrated target engagement [VERIFIED — Evommune press release 2026-06-29, quoted].

Read that carefully, because it narrows the possibilities. The drug reached the target and did nothing useful in the disease where the target’s case is strongest. That leaves three live explanations, and only one of them is good for September:

1. The mechanism does not carry enough weight clinically, anywhere. Then AD fails too.
2. CSU was the wrong disease — its dominant driver is autoimmune IgE biology, not MRGPRX2, and AD’s neuropeptide-driven itch is genuinely the better target. Then AD can succeed. This is the company’s implicit position and it is not unreasonable, but it was also not the prior anyone held before the CSU result.
3. Target coverage was insufficient at the doses tested. The company says target engagement was demonstrated, which argues against this — and the AD trial uses the same molecule, so a coverage problem would not be fixed by changing indication.

There is a fourth consideration that is not about biology at all. This trial randomises 120 patients across four arms — about 30 per arm. That is small for EASI percentage change, where placebo responses in atopic dermatitis routinely run 20–35%. A trial that size can only detect a large effect. A real but modest benefit would very likely miss. This is a design risk sitting on top of the biological one, and it points the same way.

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

Trials

Feasibility matrix

Resourcing sufficiency. Not a constraint on this readout, and not a constraint on the next decision either. The company holds $288.0M of cash, equivalents and investments and says it has “cash through 2028” (../company.md C.3). The trial is fully enrolled and the database is closing. Every cost between here and the September readout is already sunk. Whether the company can fund a Phase 3 in atopic dermatitis alone is a different and much harder question, answered in B.3b: probably not, without either a partner or a raise.

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

A.3a Target label. An oral treatment for adults with moderate-to-severe atopic dermatitis whose disease is not adequately controlled by topical therapies.

A.3b Target product profile

A.3c Strategic Go/No-Go questions. The asset’s next decision, if this readout is positive, is whether to run a registrational Phase 3, so the pre-Phase-III set is the one filled.

A.3d Regulatory designations. None disclosed for EVO756 in atopic dermatitis. No Fast Track, Breakthrough Therapy, Orphan Drug or other designation appears in the company’s press feed, in the trial registry, or in any source this sweep reached [VERIFIED — search of BPIQ fetch_company_press_releases 2026-08-13 (141 items) and ClinicalTrials.gov NCT07150845 found none]. This is unsurprising rather than alarming: atopic dermatitis is a large, well-served indication, so none of the expedited routes would be expected to apply.

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.

The honest read of this matrix. EVO756’s plausible path to value runs through the safety and convenience rows, not the efficacy row. Nothing in the evidence base suggests it will match dupilumab on skin clearance. What it could offer is a pill with no boxed warning and no injection — which is worth real money, but only once the efficacy row clears its minimum threshold. Everything above the minimum-threshold column is currently unearned.

A.5 Evidence quality and endpoints

Endpoints

A.5b Key opinion leaders.

Panel as of. 2026-08-13.

Investigators

No investigators were returned, and the absence is itself sourced rather than assumed. CT.gov search_investigators returned {"count": 0, "trials_analyzed": 0, "investigators": []}, and get_trial_details on NCT07150845 confirms it: the record lists 29 sites — mostly private dermatology research practices such as Saguaro Dermatology, Skin Care Research and Metropolis Dermatology — with no overall officials and no site contacts named. This is normal for a commercially run dermatology Phase 2 at private research sites, and it means there is no named principal investigator on this program to assess.

Independent voices

No qualifying independent voice was found, and the search that failed to find one is on record. The MRGPRX2 field has plenty of named, published commentators, and every one this sweep identified was disqualified for a stated reason:

• Joshua Wollam, Veena Viswanath, Marcus Boehm and colleagues — the most directly relevant clinical-stage commentary on MRGPRX2 antagonism, but all are Escient Pharmaceuticals employees, i.e. the direct competitor [VERIFIED — author affiliations, J Allergy Clin Immunol 2024, [DOI](https://doi.org/10.1016/j.jaci.2024.07.002). According to PubMed.]
• Martin Metz and Stefan Frischbutter (Charité, Berlin) and Nicolas Gaudenzio (Toulouse; also Genoskin SAS) — academic co-authors on that same competitor-sponsored paper, so they carry a disclosed relationship with a rival sponsor [VERIFIED — same paper. According to PubMed.]
• Mukesh Kumar, Billy K C Chow and colleagues (University of Hong Kong) — authors of the closest published support for MRGPRX2 blockade in atopic dermatitis, and they hold no disclosed relationship with Evommune. They are excluded on a different ground: they are developing their own MRGPRX2 antagonist, GE1111, which is a competing academic interest, and their commentary is on mouse data rather than on this program’s clinical endpoint [VERIFIED — Front Immunol 2024, [DOI](https://doi.org/10.3389/fimmu.2024.1406438). According to PubMed.]
• Bryan L Roth, Xinzhong Dong, Brian S Kim, Hydar Ali — leading independent academics on the target, none of whom has commented on this program or on this endpoint at all.

Recording an empty table rather than promoting one of these people to “independent” is the point of the exercise. The absence of a disclosure is not evidence of no conflict, and a commentator on the target is not a commentator on this program’s endpoint.

Judgement

Dissent

(Empty. Endpoint supported is UNKNOWN rather than SUPPORTIVE_CONTESTED, so no dissent row is owed, and inventing a disagreement nobody has voiced would be worse than an empty table.)

B. Commercial assessment

B.0 Current treatment algorithm

What an adult with moderate-to-severe atopic dermatitis receives today, in order:

1. Emollients and topical corticosteroids. Moisturisers plus steroid creams. Most patients never go further. Topical calcineurin inhibitors and topical JAK inhibitors sit alongside.
2. Phototherapy, where it is available and the patient can attend.
3. Biologic injections — the current anchor of care. Dupilumab, which blocks the shared receptor for interleukin-4 and interleukin-13, is the dominant agent. Tralokinumab and lebrikizumab block interleukin-13; nemolizumab blocks interleukin-31, the itch cytokine.
4. Oral JAK inhibitors. Upadacitinib and abrocitinib. Highly effective and fast, and carrying boxed warnings for serious infection, mortality, malignancy, major cardiovascular events and thrombosis. Many patients and prescribers avoid them for exactly that reason.
5. Older systemic immunosuppressants — ciclosporin, methotrexate — still used where access to the above is limited.

Where EVO756 would fit. As an oral option at step 4, positioned ahead of the JAK inhibitors on safety rather than behind the biologics on efficacy. That placement only works if it is meaningfully effective; a weak oral in a market with dupilumab in it has no natural home. It is additive rather than displacing at first — a thing to try before escalating to an injection or a JAK.

B.1 Competitive landscape and positioning

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

It wins on being alone. First-in-class mechanism, no clinical competitor left standing in the class, a clean safety record across three human trials, and an oral route into a market whose fastest- growing segment is oral. If EVO756 works even moderately in atopic dermatitis, it has a market largely to itself in a way very few Phase 2b assets do.

The single fact the thesis rests on is not any of that. It is whether blocking MRGPRX2 does enough, in a disease where mast cells are a contributor rather than the driver, to move EASI — after the same molecule failed to do so in the disease where mast cells are the driver. Every competitive advantage listed above is worth nothing if that answer is no, and every one of them is worth a great deal if it is yes. That is what makes this a genuinely binary event rather than a gradual re-rating.

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. Both halves apply here, and the second half is the one currently being tested.

One more competitive note that cuts the other way. Incyte paid roughly $750M for Escient in April 2024 substantially for this target class [WEB ESTIMATE — Incyte investor release via search, 2026-08-13]. That is evidence a large, sophisticated buyer believed in MRGPRX2 blockade. It is also now evidence of how that bet has gone: the lead asset is paused on toxicology and there has been no positive clinical validation of the class from either sponsor.

B.2 Addressable market

Launch markets: the United States and the EU5. Japan, Greater China and several other Asian countries are already out-licensed to Maruho and are not Evommune’s to sell (see B.3b), so they are excluded from every figure below. Ignoring that exclusion would overstate the opportunity by a large margin, which is why it is stated here rather than buried.

The overall market is large and growing: roughly $29–30 billion globally by 2030 across several independent forecasts, with the oral route the fastest-growing segment [WEB ESTIMATE — Grand View Research, Precedence Research and Spherical Insights via search, 2026-08-13. Market-sizing reports of this kind are consistent with each other partly because they cite each other; treated as an order of magnitude, not a figure.]

B.3 Value and feasibility

B.3a Expected peak sales. Built bottom-up as eligible patients × annual net price × peak penetration. Every figure below is conditional on approval, excludes Japan and Greater Asia (out-licensed to Maruho), and excludes the migraine indication entirely.

The base of the calculation is the systemically treated moderate-to-severe adult AD population in the US and EU5, taken as 750,000 patients [UNVERIFIED — modelled, the midpoint of the 600,000–900,000 range in B.2]. Net price is taken at $18,000–25,000 a year, below the oral JAK inhibitors’ list prices to reflect gross-to-net and a positioning that competes on safety rather than on efficacy [UNVERIFIED — modelled; no net price for any AD oral was verifiable this sweep].

Which input is weakest, named explicitly: peak penetration. The patient base and the price are both anchored to observable market structure and vary by less than a factor of two across the scenarios. Penetration varies by five times and rests entirely on an efficacy result that does not yet exist. A reader who disagrees with this table should disagree with the share column.

What this is worth against the company. Enterprise value is about $229M (../company.md C.4). The base case is roughly 1.8× enterprise value in annual peak sales, and the high case roughly 4×. That is a genuinely attractive ratio — and it is the ratio you would expect on an asset the market has assigned a low probability of working.

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Partly. The trial is well designed in kind — randomised, double-blind, placebo-controlled, dose-ranging, on the standard endpoint in the standard population — so a positive result would genuinely support the efficacy claim. It does not support the claims that matter most commercially, because at n≈30 per arm it cannot reliably characterise the dose-response the Phase 3 would need, and it runs only to week 12 when the registrational timepoint is week 16.
2. Will the identified risks affect the target product profile? The dominant risk — that the mechanism does not carry enough clinical weight — does not affect the profile, it deletes it. There is no version of EVO756 in atopic dermatitis that survives a second failed Phase 2b. The secondary risk, underpowering, threatens a different failure: a real but modest drug being discarded on a trial too small to see it.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? On safety and on mechanism, yes — and uniquely so, with the only rival paused on toxicology. But differentiation without demonstrated efficacy is not a commercial asset. The migraine program and EVO301 are what would remain.

Readout

Sources

The modelled estimate. The registry’s primary completion date is 2026-07. Adding the stated default lag to database lock and analysis of two to four months gives a topline window of 2026-09 to 2026-11. data/benchmarks/readout-lag.json holds no matching entry for a dermatology Phase 2b, so the default lag is used and the estimate is tagged [UNVERIFIED — modelled, default lag] rather than benchmarked. The company’s September guidance sits at the earliest edge of this modelled range, which is a mild caution rather than a contradiction: guidance at the optimistic end of what the arithmetic allows is guidance with no slack in it.

Window

Basis. The company named a specific month, September 2026, on 2026-08-06, for a trial it describes as fully enrolled, whose registry primary completion has already passed (2026-07). Precision is MONTH because no source names a day — not “Q3 2026”, and not “September 2026” either. Confidence is HIGH because three independent sources agree in direction and the guidance has tightened rather than slipped at every update, which is the pattern of a company that knows its own database-lock date.

Disagreement. CONSISTENT. No source contradicts another. BPIQ’s “Q3 2026” is simply coarser than the company’s “September 2026” and contains it; the registry’s 2026-07 primary completion sits correctly before both; the modelled 2026-09 to 2026-11 range contains the guided month at its early edge. Nothing here needed to be reconciled or chosen between.

Date slippage. Zero slips, and the guidance has tightened at every update. This is a finding in its own right and it is unusual in this corpus.

H2 2026 → Q3 2026 → September 2026 is a monotonic narrowing across five months, including a restatement on the worst day the company has had. Nothing was pushed out.

Attribution

Status. CLEAN — computed, not authored, by lib/clustering.mjs’s attributionFor over this ticker’s full pipeline for bpiq_drug_id 20071, with CATALYST_CLUSTER_MIN_MONTHS = 6.

Conflicts

(Empty, exactly as CLEAN requires.)

Why it comes out clean, for a reader who wants to check the arithmetic. Only two rows on this ticker carry has_catalyst: true: this program (20071) and EVO756 in migraine (20081). This program contributes its own readout.window of 2026-09-01 to 2026-09-30. The migraine row carries catalyst_date 2027-12-31 with catalyst_date_text “2027” and catalyst_date_is_exact: false, so it is widened to a whole half-year, 2027-07-01 to 2027-12-31. The gap between the two windows is about nine months, comfortably beyond the six-month bar. The other two pipeline rows — EVO301 (19737) and the discontinued CSU program (20079) — both carry has_catalyst: false and contribute no window at all.

Note. (Not required. CLEAN and INDETERMINATE owe no note, and this is CLEAN.)

Market and timing for this event

• Plain takeaway. The market has already marked this down hard — the shares are 57% below their 52-week high and sit at 16.6% of their 52-week range, having fallen 40% in a session on the sister trial’s failure. Short interest has rebuilt to 12.35% of shares outstanding into the readout, and two insiders bought in the open market below today’s price after the failure. This is a contested, two-sided setup rather than a crowded one.
• Months to this catalyst. 0.6 months to readout.window.earliest (2026-09-01), and 1.6 months to readout.window.latest (2026-09-30). Precision is MONTH, so the event is known to within a month but not to a day.
• Expected move around this event. Not readable from the options chain, and no point estimate is offered. ../company.md C.6 records the chain as unusable: at the $15.00 strike on the 2026-09-18 expiry — the strike nearest the $14.25 spot — both the call and the put are bid $0.00 with zero open interest, so no at-the-money straddle can be constructed at all. The bracket this document works with comes instead from this ticker’s own history (C.7): a positive dermatology Phase 2 readout was worth about +71% in a session, and a failed Phase 2b on this molecule about −40% close to close.
• Nearest comparable past reaction. Two rows of ../company.md C.7 are relevant and they are comparable in different ways, so both are named rather than one chosen. For the downside, 2026-06-29 is the closest analogue of any kind: the same molecule, the same Phase 2b dose-ranging design, a genuinely blind binary, −40.2% close to close. For the upside, 2026-02-10 is closest by disease: a positive placebo-controlled Phase 2 readout in atopic dermatitis on this ticker, +70.9% close to close. The upside analogue is the weaker of the two — it was a different molecule (EVO301), and it landed while the urticaria program was still alive and still the lead story, so the company it re-rated was a different company from today’s.
• Materiality. Dominant, near-term, as recorded in ../company.md C.2. This is the next binary event on the calendar and it is on the lead molecule; after the CSU failure, EVO756’s entire remaining dermatology value rests here. The stock-direction call below is consistent with that in the sense that matters: both scenario ranges are wide and far apart. What keeps the miss scenario off zero is that “dominant” is not “sole” — $288.0M of cash, the migraine program and EVO301’s positive Phase 2a all survive a failure.
• Date slippage. Zero slips, and the guidance tightened at every one of four updates (Readout, above). Worth stating plainly because zero is a finding, not an absence of one: this is a company that has told the market the same thing, more precisely each time, including on the day it announced a failure.

Spot. $14.25, the price cache’s close on 2026-08-13, cited from ../company.md C.4. BPIQ’s intraday last price the same day was $14.32. The cache close is used because a run-up settlement must read the committed cache rather than a fresh fetch, and entry and settlement measured from two different sources would not be comparable.

Scenario prices

Expected value. Applying the modelled 30% probability to the midpoint of each range: 0.30 × $25.50 + 0.70 × $9.75 = $14.48, which is +1.6% against the $14.25 spot. Method: the probability applied to the midpoint of each scenario range. This is arithmetic, not advice, and it is not a price target. What it says is worth stating plainly: at a 30% probability, today’s price is very close to fair for the two outcomes described. The market and this analysis disagree about almost nothing on level — the disagreement, if any, would be about the width of the distribution.

Run-up

date_confidence scores 60, well clear of the zero floor that rule 39 would use to withhold this call, so a run-up call is made.

Entry and exit

exit is null at lock time and is recorded as such rather than guessed: the price cache ends 2026-08-13 and the window does not open until 2026-09-01, so lib/runup.mjs’s resolveExit has no trading history to count back from and correctly returns null. It resolves once the cache reaches the window.

Note the shortness of this trade. Entry to exit is roughly eight trading days. This is not a long accumulation into a distant catalyst; it is a brief position taken because the readout is already almost here. The predicted move is small for exactly that reason.

Predicted move and peak

Priority score drivers

Transcribed from lib/runup.mjs’s scoreDriver, not hand-computed.

Priority score. Priority score 39 · formula_version 1.0.0

A middling score, and the shape of it is worth reading rather than the number alone: the two drivers doing the work are priced-in-ness (83) and value uplift (70), and the one holding it back is the 30% probability of a positive outcome. This is the profile of a cheap lottery ticket on a well-dated event, not of a high-conviction run-up.

Settlement. 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. No position on direction; if anything, a small, short, explicitly speculative run-up position sized to be lost.

Why. The evidence points one way and the price already reflects a good deal of it. The same molecule failed the same kind of trial, at every dose, in the disease where its mechanism has the strongest case, and this trial asks the mechanism to do more in a disease where it plausibly matters less — with about 30 patients per arm against placebo responses that routinely run 20–35%. Against that, the shares are 57% off their high at 1.36× cash per share, two insiders bought in the open market below today’s price after the failure, funds added through July and August, and the probability-weighted arithmetic lands within 2% of spot. There is no edge on direction here, which is a more honest answer than a view dressed up from the same facts.

What would change this. A single observable: the September topline. A statistically significant separation from placebo on EASI percentage change at week 12 at any dose reverses the central argument of this document outright, because it would be the first clinical evidence that blocking MRGPRX2 does something useful in any disease. Short of the readout, the thing that would move the view most is a clean readout of the pruritus NRS secondary — non-histaminergic itch is where this mechanism should show up first, so a large itch effect alongside a marginal EASI result would suggest a real drug in an underpowered trial rather than an inert one.

What to watch.

• Now through 2026-08-31 — any company statement narrowing “September 2026” to a day, or naming a congress. Neither exists today, and either would raise readout.precision from MONTH.
• 2026-08-14 and 2026-08-31 — the next two FINRA short-interest settlements (published roughly two weeks late). Short interest has grown 45% in three months; whether it keeps building into the print or starts covering is the cleanest read on positioning available.
• Ongoing — EDGAR Form 4 filings. The two July open-market purchases are the only discretionary insider trades on the record; a third, or a discretionary sale, would be genuinely informative. Note that the BPIQ insider connector will not show these — it returned zero rows across three attempts against 64 filings in EDGAR.
• 2026-09-01 to 2026-09-30 — the topline release itself, expected as a Business Wire release and a Form 8-K on the same day.
• On the day — read the pruritus NRS and EASI-75 secondaries, not just the primary. A trial this small can miss its primary and still carry a signal, and the company’s next decision will be made on the whole package.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): miss — probability 30%, band 20–45% [UNVERIFIED — modelled]. The probability that the settlement definition below is met. Built from: the same molecule’s failure at every dose in the mechanistically stronger indication (the dominant term); a disease whose established drivers sit outside this mechanism; ~30 patients per arm against AD placebo responses of 20–35%; no human efficacy evidence for EVO756 in AD of any kind; and no peer-reviewed publication of this molecule’s clinical data anywhere. Set against those: genuine preclinical validation of MRGPRX2 in AD models, demonstrated target engagement, a clean safety record, and a company that saw the unblinded CSU result and chose to keep this trial running. No third party has published a probability on this specific event that this sweep could find.
• Stock-direction (which way do the shares move?): no-edge — confidence medium — window 2026-09-01 to 2026-10-15, basis: readout.window (2026-09-01 to 2026-09-30) plus a fortnight for the move to settle. Cites the dominant, near-term materiality recorded in ../company.md C.2, and is consistent with it: both scenario ranges are wide and far apart, as a dominant catalyst requires, and the call declines direction rather than magnitude.
• Scenario prices: positive $21.00–$30.00 · miss $8.50–$11.00
• Expected value: $14.48, +1.6% against spot $14.25. The no-edge call and this number agree: +1.6% is immaterial, so no explanatory note is owed.
• Run-up: entry $14.25 on 2026-08-13, exit rule T-5 trading days before readout.window.earliest — predicted move +2–15%, predicted peak +5–20% around 2026-09 — priority score 39, formula_version 1.0.0
• Settles on the publication of top-line Phase 2b results for EVO756 in moderate-to-severe atopic dermatitis (NCT07150845), by Evommune press release or Form 8-K. Positive means the company reports a statistically significant improvement versus placebo (p < 0.05) in the percentage change from baseline in EASI at week 12, at one or more of the three EVO756 dose regimens. A company statement that the trial did not meet its primary endpoint at any dose, or that discontinues or deprioritises EVO756 in atopic dermatitis without claiming the primary endpoint was met, settles as a miss.
• Locked: yes · Settled: no

Program data-quality flags

• PubMed’s short-code collision fired on this program in its purest observed form. A search for EVO756 returns two records, one of which is a 2009 paper on hybrid speciation in butterflies whose publisher item identifier is literally the string EVO756 [DOI](https://doi.org/10.1111/j.1558-5646.2009.00756.x). A hit count taken at face value would have doubled this molecule’s apparent literature. The real count of primary publications on EVO756 is zero, and that zero is a finding in A.5, not an absence.
• The PubMed null-authors bug did NOT fire this sweep. All eight articles returned complete author lists with affiliations, which is what made the A.5b independence assessment possible at all. Recorded because the bug is documented as expensive precisely when it does fire.
• Open Targets search_entities required three attempts. The first two returned the standing platform throttle, verbatim: Rate limit exceeded for client: global. The third succeeded. Recorded CALLED rather than BLOCKED because the documented retry policy resolved it — but the block is transient rather than gone, and a sweep that stopped after two attempts would have recorded a false BLOCKED.
• ChEMBL returned a legitimate empty, not the HTTP 500 recorded elsewhere. compound_search with name=EVO756 returned {"count": 0, "total": 0, "compounds": []}. EVO756 is simply not in ChEMBL v34. The cost is stated in section 0: no independent selectivity data, so the “highly selective” claim rests on the company alone, and the question of whether the CSU failure could reflect insufficient target coverage cannot be answered from this sweep.
• One competitor fact is a WEB ESTIMATE where a primary source was wanted. The report that Incyte paused and then halted EP262/INCB000262 over preclinical toxicology findings comes from a search summary; the Fierce Biotech article itself returned HTTP 403 Forbidden to a direct fetch and could not be read. It is tagged as an estimate throughout B.1 and A.2 rather than promoted to verified. Nothing in the prediction depends on it — it strengthens EVO756’s relative safety position and weakens the class’s clinical validation, and those roughly cancel.
• The composition-of-matter patent term for EVO756 could not be established. Recorded as an open gap in B.1 and B.2 rather than estimated. It does not touch the September readout or any figure in the locked prediction.
• The upstream royalty rate owed to Dermira (now Eli Lilly) is not public. The existence and exclusivity of the December 2020 worldwide licence are established; the economics are not. This makes every eNPV figure in B.3b a range rather than a point, which rule 10 would require in any case.
• One BPIQ field on this row is stale relative to BPIQ’s own note. catalyst_date_text reads “Q3 2026” while the same row’s note field records the company’s 2026-08-06 update to “September 2026”. The readout block is built on the company’s own release rather than on either BPIQ field, so nothing downstream inherits the staleness — but a reader taking catalyst_date_text at face value would carry a coarser date than the company has actually given.
• No source conflict was left unresolved at the program tier. All five readout sources agree in direction; disagreement is CONSISTENT rather than UNRESOLVED. The one open conflict on this ticker is the company-level runway disagreement, recorded in ../company.md C.3 and C.8.