RAPP / rap-219-bipolar-mania — RAP-219 for acute manic episodes in bipolar I disorder

Program analysis · bpiq_drug_id 18871 · prepared 2026-08 · USD · framework v5.10.3 · Coverage: CLEARED

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

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

Glossary

Executive summary

• What it is (one sentence): A once-daily tablet that damps down a specific subset of the brain’s main excitatory switches — the ones found in the forebrain and not in the balance and breathing centres — tested against placebo in 253 hospitalised patients having a manic episode, to see whether their mania scores fall further than on a dummy tablet over three weeks.
• The event and when (as disclosed): Topline results from the Phase 2 bipolar mania study NCT07046494. The company said on 2026-08-05 that topline results are “expected in October 2026” [VERIFIED — Rapport second-quarter 2026 release, 2026-08-05]. No source names a day. The readout window used throughout this document is 2026-10-01 to 2026-11-30, most likely the second half of October.
• The main reason it could work: The drug is proven to reach its target in the living human brain and to do something clinically real with it. In focal epilepsy it produced a 77.8% median reduction in clinical seizures against baseline over eight weeks, with 24% of patients becoming seizure-free [VERIFIED — Rapport topline release, 2025-09-08]. Acute mania is also one of the more tractable psychiatric endpoints: approved drugs beat placebo on the YMRS reliably and by margins that are not subtle. And this trial is large for a proof of concept — 253 patients, double-blind, placebo- controlled — so it is not underpowered for a moderate effect.
• The main risk: Every drug approved for acute mania either blocks dopamine or is a classical mood stabiliser. No novel non-dopaminergic mechanism has ever succeeded here. The one recent attempt — Biohaven’s BHV-7000, a potassium-channel opener, 274 patients, 21 days, inpatient, the same YMRS endpoint at the same timepoint — missed outright, and Biohaven abandoned psychiatry [VERIFIED — Psychiatric Times and Clinical Trials Arena, 2026-03]. Thirteen of RAP-219’s twenty trial sites also ran that failed study [VERIFIED — ClinicalTrials.gov location lists for NCT07046494 and NCT06419582, read 2026-08-19]. High placebo response at professional psychiatric research sites is the single commonest cause of failure in this indication, and this trial is running at the same sites where it has just been demonstrated.
• What it means for the stock: Less than the price action suggests. The shares sit at 98.4% of their 52-week range and at an all-time high, up 3.5× from the low (../company.md C.4), and most of that value is a bet on the epilepsy Phase 3 program that reports in 2029, not on this readout. The honest call is no-edge: the expected value of the two scenarios lands within about one percent of the current price. The run-up call is more informative than the direction call, and even it scores a low priority.

0. Program-tier coverage — CLEARED

Every mandatory program-tier row was called. Two are recorded as failures with their verbatim errors, and both are named again where the missing data would have been used. Company-tier coverage is in ../company.md C.0 — it is not repeated here.

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

What the drug is. RAP-219 is a small molecule taken as a tablet once a day. It works on the AMPA receptor, which is the brain’s main fast excitatory switch: the messenger glutamate lands on it and the nerve cell fires [VERIFIED — Bialer et al., *Epilepsia* 2024, [doi:10.1111/epi.18056](https://doi.org/10.1111/epi.18056), via PubMed]. Drugs that block AMPA receptors everywhere in the brain do stop seizures — perampanel is an approved example — but they also cause dizziness, unsteadiness and psychiatric side effects, because the same receptors run the cerebellum and the brainstem.

The selectivity idea. RAP-219 does not target the AMPA receptor itself. It targets TARP γ-8, a helper protein that sits alongside AMPA receptors but is concentrated in the forebrain — the hippocampus and cortex — and is scarce in the cerebellum and brainstem [VERIFIED — Bialer et al., *Epilepsia* 2024, [doi:10.1111/epi.18056](https://doi.org/10.1111/epi.18056)]. By binding TARP γ-8 rather than the receptor, the drug damps down forebrain AMPA signalling and largely leaves the rest alone. It is a negative allosteric modulator: it turns the response down rather than switching it off. The company describes it as potentially first-in-class [VERIFIED — Rapport press release, 2026-03-09].

How well the target is validated — and this is where the two diseases part company.

For epilepsy, validation is real and human. RAP-219 reached its intended receptor occupancy in living human brains, measured by brain imaging in Phase 1 [VERIFIED — BPIQ historical catalyst row, 2025-01-09], and then produced a 77.8% median reduction in clinical seizures against each patient’s own baseline over eight weeks, with 24% becoming seizure-free, in the Phase 2a study [VERIFIED — Rapport topline release, 2025-09-08; the trial also met its pre-specified primary endpoint on device-recorded long episodes]. Independent academics reviewing the epilepsy pipeline list it among treatments with reported seizure-outcome data [VERIFIED — Bialer et al., *Epilepsia* 2026, [doi:10.1002/epi.70346](https://doi.org/10.1002/epi.70346), via PubMed].

For bipolar mania, validation is inferential. The argument runs: excess glutamate signalling in forebrain circuits contributes to mania; damping forebrain AMPA receptors should therefore reduce manic symptoms. That is a plausible hypothesis and it is not a demonstrated one. There is no human efficacy data for RAP-219 in any psychiatric condition, no publication on the subject, and — because Open Targets was BLOCKED (section 0) — no independent human genetic evidence in this document either way. [UNVERIFIED]

The exact scientific step this readout must prove. That damping forebrain AMPA receptors reduces manic symptoms more than placebo does, measured on the YMRS at three weeks, in patients selected by clinical diagnosis alone. Every fact supporting the molecule so far is about stopping seizures in an epileptic brain, measured against each patient’s own untreated baseline in an open-label study where everyone knew they were taking the drug. This readout asks a different question in a different disease against a different comparator.

The honest scientific risk. It is not that the drug does nothing. It is that every approved treatment for acute mania works through dopamine blockade or is a classical mood stabiliser — lithium, valproate, carbamazepine — and no novel mechanism has ever displaced them. Acute mania has consumed a long series of mechanistically interesting drugs. The most recent, and the closest to this one in design, is Biohaven’s BHV-7000: a potassium-channel opener, 274 patients, 21 days, inpatient, YMRS at day 21 — the same trial in all but the molecule — which missed its primary endpoint, after which Biohaven stopped running psychiatric trials altogether [VERIFIED — Psychiatric Times and Clinical Trials Arena reports of the 2026-03-03 announcement; trial design from ClinicalTrials.gov NCT06419582].

There is a second, more technical risk that deserves naming. RAP-219 has a half-life of about 22 days [VERIFIED — Rapport press release, 2026-04-21]. A drug that slow reaches steady concentrations over months, not weeks. In a 21-day trial, that means blood levels may still be climbing when the primary endpoint is measured — unless a loading regimen is used. No source read this sweep discloses the dose or the regimen. The registry says only “RAP-219 tablets administered orally, once daily for 21 days” [VERIFIED — ClinicalTrials.gov API v2, NCT07046494 arms and interventions, read 2026-08-19]. This cuts both ways and it is honest to say so: the same long half-life is why the epilepsy effect persisted for weeks after dosing stopped, which suggests exposure builds usefully fast. But the question cannot be settled from what is public.

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

Trials

The site overlap, which is the most specific fact in this document. Comparing the two location lists directly, 13 of RAP-219’s 20 sites are the same research organisation in the same city as a site that ran the failed BHV-7000 trial: Rogers (Arkansas), Lemon Grove and Orange (California), Hollywood, Miami Lakes and West Palm Beach (Florida), Decatur (Georgia), Chicago (Illinois), Gaithersburg (Maryland), St Louis (Missouri), Staten Island (New York), and Austin and Richardson (Texas) [VERIFIED — ClinicalTrials.gov location lists for NCT07046494 and NCT06419582, both read 2026-08-19].

Why this matters rather than being a curiosity: acute mania trials do not usually fail because the drug is inert, they fail because the placebo arm improves too much. Hospitalised patients are removed from stress, sleep, eat and are watched around the clock, and all of that improves a YMRS score on its own. Specialist psychiatric research sites that recruit for study after study are where that effect is largest. Whatever placebo response sank BHV-7000 was generated substantially at these same thirteen sites, and it is a property of the sites and the population as much as of the drug. [UNVERIFIED — the causal reading is the analyst's own; the site overlap itself is verified]

The enrolment increase and the changed analysis plan. On 2026-05-07 the company moved topline guidance from the first half of 2027 to the fourth quarter of 2026 and said it had “increased target enrolment” and modified the trial’s statistical analysis plan, adding that the higher enrolment might let the study serve as confirmatory evidence of effectiveness for a future approval application. It repeated all of this on 2026-08-05 [VERIFIED — BPIQ note field on drug id 18871, and the second-quarter 2026 release]. Two readings are available and both are legitimate. The generous one: the sponsor saw blinded aggregate data suggesting the trial could carry a regulatory burden and invested to make it do so. The cautious one: a mid-trial power increase is what a sponsor does when a blinded look shows the variance is higher, or the placebo arm better, than the original plan assumed. Nothing public separates them. [UNVERIFIED]

Feasibility matrix

Resourcing sufficiency. Comfortable, and not the binding constraint. The company held $436.1M of cash and short-term investments at 2026-06-30 against a recomputed burn of $13.57M a month, with no debt — see ../company.md C.3. The trial this readout comes from is already fully enrolled and dosed, so its remaining cost is analysis. The real resourcing question sits elsewhere: two Phase 3 epilepsy trials totalling 645 patients began in 2026 and run to 2029, which is what will consume the balance sheet, and the recomputed runway (to about 2029-03) already falls short of the company’s stated one (into the second half of 2029) before that ramp is accounted for. A positive bipolar readout would add a third development program to fund, not relieve the pressure.

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

A.3a Target label. Acute treatment of manic episodes, with or without mixed features, associated with bipolar I disorder, in adults [VERIFIED — ClinicalTrials.gov NCT07046494 title].

A.3b Target product profile

A.3c Strategic Go/No-Go questions. This asset’s next decision on this indication is whether to take bipolar mania into a registration program, so the pre-Phase-III set is the one that applies.

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

A.3d Regulatory designations. None disclosed for this indication. No breakthrough therapy, fast track, orphan or priority designation is named in any source read this sweep for RAP-219 in bipolar disorder. The one regulatory fact on record is a negative one from a different indication: the application to begin human testing in diabetic peripheral neuropathic pain was placed on clinical hold, the hold was later lifted, and the company has deferred the program [VERIFIED — BPIQ note field on drug id 19422]. The nearest thing to a designation here is the company’s statement that this Phase 2 could potentially count as confirmatory evidence of effectiveness — that is a claim about how the trial might be used, not a designation granted by anyone [VERIFIED — Rapport second-quarter 2026 release, 2026-08-05].

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 transfers to acute mania, where the comparators are oral generics and there is no survival endpoint — recorded because the template carries them, and flagged as inapplicable here rather than silently borrowed.

A.5 Evidence quality and endpoints

Endpoints

A.5b Key opinion leaders.

Panel as of. 2026-08-19 — the date both searches below were run.

Investigators

Exactly one person is disclosed. search_investigators returned nothing (section 0); the overall official was recovered from the trial record itself. Twenty sites are listed and no site investigator is named for any of them, which is ordinary for a registry record but means this table is one row where a fuller disclosure would have given twenty.

Independent voices

Empty, and the emptiness is the finding. Two searches were run. PubMed on RAP-219 returned three articles; the two relevant ones are conference-summary reviews in Epilepsia whose authors — Bialer, Johannessen Landmark, Koepp, Perucca, Perucca, Tomson, White and Wirrell — are independent academics with no disclosed Rapport relationship. But they write about RAP-219 as a drug-resistant focal epilepsy candidate and say nothing whatever about bipolar disorder or the YMRS. They are therefore not commentators on this program’s endpoint and recording them here would misrepresent what they said. search_investigators returned nothing.

Not one named independent voice has commented publicly on whether damping forebrain AMPA receptors should reduce manic symptoms. No row is written rather than a weak one being manufactured.

Judgement

Dissent

Empty. Endpoint supported is UNKNOWN rather than SUPPORTIVE_CONTESTED, so no dissent row is required and none is invented.

B. Commercial assessment

B.0 Current treatment algorithm

What a patient in a manic episode receives today, in order:

1. Admission. A severe manic episode is usually treated in hospital, because insight is lost and the behaviour is dangerous. Admission itself improves symptoms — regular sleep, no stress, constant supervision — which is the reason placebo arms in these trials improve as much as they do.
2. First line: one drug, started immediately. Either a dopamine-blocking antipsychotic (olanzapine, risperidone, quetiapine, aripiprazole, asenapine, cariprazine, ziprasidone, and since February 2026 milsaperidone), or lithium, or valproate. All work. The choice is made mostly on side effects, speed and what the patient has tolerated before.
3. Second line: combine. An antipsychotic plus lithium or valproate, if one drug is not enough.
4. Third line: electroconvulsive therapy, or clozapine, for episodes that resist everything else.
5. After the episode: maintenance, to stop the next one. This is where most of the money in bipolar disorder is spent, and it is a different — and much larger — commercial question than the one this trial answers.

Where RAP-219 would fit: step 2, as an alternative first-line agent, for the patient who cannot tolerate dopamine blockade. It is not competing on how fast mania resolves; the existing drugs are already good at that. It is competing on what the patient has to accept in exchange — weight gain, metabolic disease, restlessness, stiffness, raised prolactin, and the risk of a permanent movement disorder. That is a real clinical gap and it is also a narrower commercial position than “a new treatment for mania” sounds, because it starts as a second choice for a subset of patients.

The commercial consequence cuts both ways, and it decides the valuation in B.3a. An acute-mania label alone is worth very little: a three-week course of any drug, however good, generates a few hundred dollars per patient per episode. The value lies entirely in whether an acute approval opens the door to maintenance use, where patients take the drug for years. Every peak-sales figure in B.3a is built on that assumption, and it is the assumption most likely to be wrong.

B.1 Competitive landscape and positioning

Named competitors

Where this asset wins, and the single fact the thesis rests on. It wins on tolerability, and only on tolerability. Nobody needs a drug that resolves mania faster than olanzapine does; people need one that resolves it without the weight gain, the akathisia and the long-term movement risk. If RAP-219 separates from placebo on the YMRS with a clean tolerability profile, it becomes the first credible option for a large group of patients who currently choose between an unbearable illness and unbearable side effects.

The single fact the thesis rests on is this: that damping forebrain AMPA signalling is enough, on its own, to bring a manic episode down — without touching dopamine. Everything else follows if that is true and nothing follows if it is not. There is no partial credit here: a numerically favourable miss is a miss, and the field has no history of second chances for a non-dopaminergic mechanism.

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. The direction transfers to psychiatry; the multiples do not, and are not used in B.3a.

B.2 Addressable market

The launch markets would be the United States first — all twenty trial sites are American — then the five largest European markets and Japan. Bipolar I disorder is diagnosed and treated everywhere, so market access is a pricing question rather than an infrastructure one.

B.3 Value and feasibility

B.3a Expected peak sales.

Scope. These figures are for bipolar disorder only and exclude epilepsy entirely, which is where most of this company’s value sits. They exclude Greater China, which is licensed to Tenacia in exchange for mid-single-digit to mid-teens tiered royalties [VERIFIED — Rapport press release, 2026-03-09]. All three scenarios are conditional on eventual approval, which requires this readout to succeed and then a full registration program to succeed after it.

The structural assumption, stated first because it drives everything. An acute-mania label on its own is nearly worthless commercially: three weeks of tablets per episode, a few hundred dollars. Every figure below assumes that an acute approval leads to chronic maintenance use, where patients take the drug for years, exactly as the approved antipsychotics are used today. If that does not happen, divide every number below by roughly twenty. [UNVERIFIED — modelled]

No third-party forecast was found for this program. The peak-sales web search returned only aggregate bipolar market-size reports — the market as a whole is put at roughly $5.0–5.4 billion in 2026 growing at about 2% a year [WEB ESTIMATE — Mordor Intelligence and Coherent Market Insights market reports, retrieved 2026-08-19]. That is a ceiling on the whole category, dominated by generics, and is not a forecast for this asset. It is recorded rather than converted into one.

The least defensible input is the peak share, and the second least is the price. Patient numbers can at least be bounded by epidemiology. The share depends entirely on how large and how visible the tolerability advantage turns out to be — a question this three-week trial will barely touch, since metabolic and movement side effects take months to years to appear. The price depends on a payer negotiation against generics that cost almost nothing. Neither is verifiable today and neither is presented as a point estimate. 01-rules.md rule 10 is why the whole table is a range.

Against market value. Enterprise value is $1,911.77M, recomputed from 47,978,286 filed shares at $48.94 less $436.077M of cash and short-term investments (../company.md C.4). Even the high case for bipolar is roughly 90% of the entire enterprise value; the base case is about 30% of it, before discounting for probability, time and the cost of the registration program that would be needed. Most of this company’s value is not this program, which is the same conclusion the pipeline table reaches from the other direction.

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Partly. The trial will answer the efficacy claim cleanly — a randomised, quadruple-masked, placebo-controlled study of 253 patients on the field’s standard endpoint is exactly the right instrument for that question. It will not answer the tolerability claim that the whole commercial thesis rests on, because weight gain, metabolic disease and tardive dyskinesia take months to years to appear and this trial runs for three weeks. A positive readout proves the half of the profile that matters least commercially.
2. Will the identified risks affect the target product profile? The largest one destroys it. If the placebo arm improves as much as the drug arm — the risk the BHV-7000 precedent and the shared sites make concrete — there is no product profile to discuss for this indication. The dose and half-life risk is more survivable: an under-dosed 21-day study could be repeated with a loading regimen, at the cost of a year or more.
3. If a risk cannot be mitigated, is the asset still differentiated from competitors? In epilepsy, yes, entirely. A bipolar failure would not touch the focal-epilepsy Phase 2a result, the two running Phase 3 trials, or the differentiation argument in that disease. This is the single most important structural fact about the downside here and it is why the miss scenario below is shallower than the class benchmark for a Phase 2 miss would suggest.

Readout

Sources

The modelled estimate. The arithmetic, stated so it can be argued with: the ClinicalTrials.gov primary completion date is 2026-08-05, and 01-rules.md rule 34’s stated default lag from primary completion to topline is two to four months. That gives 2026-10-05 to 2026-12-05. data/benchmarks/readout-lag.json holds no entry matching a three-week inpatient psychiatric trial, so the default is used and the estimate is tagged [UNVERIFIED — modelled, default lag] rather than benchmarked. Two things are worth noting about how this lands. First, the company’s own guidance (October) sits at the early edge of the modelled band, not outside it — the two agree rather than conflict. Second, this trial’s endpoint is a clinician-rated scale collected at a fixed day in an inpatient setting, with no imaging to read centrally and no long-term follow-up to clean, so the fast end of the default range is the reasonable expectation here.

Window

Basis. The earliest edge is the first day of the month the company itself named in a release 14 days before the lock date, and no source suggests anything earlier is possible. The likeliest date sits in the second half of October: the company named the month, and the modelled arithmetic (primary completion 2026-08-05 plus a two-month lag) lands on 2026-10-05, so the two together point at October rather than at its first days. The latest edge extends one month past the guidance, which is the ordinary allowance for a database lock running long — deliberately not further, because the trial’s primary completion date has already passed with actual enrolment posted, which removes the usual source of slippage. Precision is MONTH because a month was named and no day was.

Confidence is HIGH, which is unusual in this corpus and is earned rather than assumed: three independent sources agree (company guidance, the BPIQ text field, and the registry’s own completion date plus the default lag), the trial has finished dosing, and the guidance has moved earlier twice and later never.

Disagreement. CONSISTENT. Every source points at October or at a band containing it. The BPIQ placeholder (2026-10-31) is the last day of the named month rather than a conflicting claim, and the modelled band’s early edge (2026-10-05) sits inside the company’s own month.

Date slippage. Two changes across five dated statements — and both moved the date earlier, which is the opposite of what this field usually records.

[VERIFIED — BPIQ fetch_company_drugs note field on drug id 18871, read 2026-08-19; the last entry corroborated directly against the company's 2026-08-05 release]

Read plainly: guidance was pulled forward by roughly six months over three months of reporting. That is an execution signal in the company’s favour and it is not a signal about the result. A trial that enrols fast reads out early whether the drug works or not.

Attribution

Status.

CONTAMINATED

Conflicts

Computed, not authored: lib/clustering.mjs’s attributionFor('RAPP', companyRecord, programs, 6, 18871) run over this ticker’s full pipeline on 2026-08-19. The third has_catalyst row (bpiq_drug_id 20353, the long-acting injectable, “2027”) widens to 2027-07-01 to 2027-12-31 and sits 213 days clear of this window, so it does not conflict.

Note. The conflicting row is the same molecule in the lead disease. bpiq_drug_id 18870 covers the initial data from the focal-epilepsy open-label extension study (NCT07219407), which the company said on 2026-08-05 it expects in the fourth quarter of 2026 — a period that fully contains this program’s own 2026-10-01 to 2026-11-30 window. The conflict is confirmed because one side of it rests on a disclosed date: this program’s readout.window carries MONTH precision.

What that means for the stock call below, stated plainly: a share-price move anywhere in October or November 2026 cannot be attributed to the bipolar readout alone. Two RAP-219 datasets are expected inside the same window, and they interact rather than merely coinciding. A positive bipolar result alongside durable extension data would compound; a bipolar miss alongside strong extension data would be partly cushioned, because the extension speaks to the epilepsy program that carries most of the equity; and a bipolar miss alongside disappointing extension data would be read as a verdict on the molecule rather than on one indication, which is much worse than either alone. The scenario ranges in the next section are written for the bipolar readout in isolation, and this is the caveat that has to travel with them. It is also why the run-up priority score’s financing_clustering_risk driver reads 100 — its worst possible value — and drags the whole score down.

Market and timing for this event

• Plain takeaway. Six weeks to the opening of a two-month window, on a stock that is already at an all-time high having tripled from its low, in front of a readout this document puts at 35% odds. The price already contains a great deal of optimism and very little insurance.
• Months to this catalyst. 1.4 months — 43 days from 2026-08-19 to readout.window.earliest of 2026-10-01, measured from the earliest edge as rule 35 requires, never from the BPIQ placeholder. readout.precision is MONTH, so the readout could land anywhere within October and the window allows into November; a reader should treat “1.4 months” as the start of a two-month period rather than a countdown to a date.
• Expected move around this event. A bracket of 32.7% to 51.3%, not a point estimate. ../company.md C.6 records the chain as low-confidence: at the 2026-11-20 $50 strike — the first expiry covering the likeliest readout date, since no October expiry exists — the call is bid $8.00 / ask $12.50 and the put bid $8.00 / ask $12.60, so buying both costs anywhere from 32.7% to 51.3% of the share price depending on where in that spread you trade. Open interest at that strike is ten contracts on one side and none on the other. This bracket sits far above the class prior and the gap is worth naming. framework/07-benchmarks.md puts a Phase 2 readout at +12% positive and −16% negative on average, a CNS readout at +16% / −25% across all phases, and a placebo-controlled design at a 29% amplitude between the two — and it records that companies under $1B in market capitalisation react 9.5× more strongly at Phase 2 than those at or above it. At $2.35 billion this ticker sits on the less sensitive side of that line. A chain implying 42% at the midpoint against a class prior in the 25–30% region is what an illiquid chain does, not a market signal. [WEB ESTIMATE — IQVIA 2024] for the class figures. One line in the chain does carry information: 1,500 contracts of open interest on the 2026-11-20 $40 put, 18% below spot, expiring three weeks after the window opens — by far the largest single line anywhere in the chain, and a real position in downside protection across this event.
• Nearest comparable past reaction. ../company.md C.7’s 2025-09-08, +39.6% intraday on the epilepsy Phase 2a topline. It is the nearest analogue by type — a proof-of-concept readout on this molecule — and it is not comparable in magnitude, for three reasons stated rather than glossed: it was the lead indication rather than a secondary one; the company had no other value driver at the time, where it now has a funded Phase 3 program; and the share count was 25% smaller against a share price around $9 before the gap, so the same dollar re-rating was a far larger percentage. The 2025-12-05 row (+1.7% on a secondary data presentation) is the closer analogue for a non-transformative release. C.7 contains no negative row at all; the nearest observations of how this equity trades bad news cluster at −7% to −8% and are all dilution or expectation-reset events rather than clinical failures.
• Materiality. MEANINGFUL, NOT DOMINANT, exactly as recorded in ../company.md C.2. A 253-patient placebo-controlled study in a second disease: success opens a large new market and validates the molecule beyond epilepsy, failure removes an option but leaves the epilepsy program untouched. The stock-direction call below is consistent with that: it is no-edge, not a directional bet, and both scenario ranges are narrower than a dominant-program readout would justify.
• Date slippage. Two changes across five dated statements, both pulling the date earlier (Readout, above). Zero delays. That is an execution finding, not a result finding.

Spot. $48.94, read 2026-08-19, cited from ../company.md C.1 and C.4. Note that the same record’s previous_close is $49.01 and that BPIQ’s own market_cap and max_52_week_position fields are both computed off that previous close rather than off this price — C.8 rows 1 and 4. Every figure below uses $48.94.

Scenario prices

Expected value. Method: the modelled 35% probability applied to the midpoint of each range — $62.00 for positive, $41.00 for miss. $0.35 × 62.00 + 0.65 × 41.00 = $48.35, which is −1.2% against the $48.94 spot. This is arithmetic, not advice, and it is not a price target. It is the direct reason the stock-direction call below is no-edge rather than down: on this document’s own probability and its own two ranges, the event is close to fairly priced.

Run-up

readout.precision is MONTH and date_confidence scores 60, so it has not floored and rule 39 does not withhold this call.

Entry and exit

exit is left null on the prediction record. Resolving the rule to a concrete date and price needs price history, and data/prices/RAPP.json does not exist — this is the ticker’s first analysis and the price connector runs on its own cadence outside an analysis run (../company.md C.0). It will resolve when the cache covers this ticker.

Predicted move and peak

Priority score drivers

Priority score. Computed by lib/runup.mjs’s priorityScore over the seven rows above, never hand-computed.

Priority score 16 · formula_version 1.0.0

Sixteen out of a hundred is a low score and the two reasons are visible in the table: the window is contaminated by a second readout on the same molecule, and the shares have already run to the top of their range. Neither is a statement about whether the drug works. This score ranks the trade, not the science.

Settlement. Null at lock time. Settled only on an explicit user request, from a confirmed primary source, against the committed price cache — which does not yet exist for this ticker.

Verdict

What I would do. Watch.

Why. The science question is genuinely open and the price question is not. RAP-219 is a real drug with proven target engagement in living human brains and a real clinical result in epilepsy, and this is a large, well-designed, fully enrolled placebo-controlled trial on the field’s standard endpoint — none of which is true of most proof-of-concept readouts in this corpus. But no non-dopaminergic mechanism has ever succeeded in acute mania, the one recent attempt failed on the identical endpoint, and thirteen of the twenty sites running this trial also ran that failed one. Set against a 35% probability, the two scenario ranges produce an expected value within about one percent of the current price, on a stock sitting at 98.4% of its 52-week range with insiders selling and a founding holder distributing. There is no edge to take here — and the window is shared with a second RAP-219 readout, so even a correct call on the science may not be a correct call on the shares.

What would change this. Three specific observables, in descending order of weight. First, the dose and regimen. If the company discloses that a loading regimen was used — or if the topline release reports drug concentrations at Week 3 near steady state — the largest technical objection in A.1 disappears and the probability should rise. Second, a lower entry. The arithmetic above is a statement about $48.94, not about the asset. A pullback into the low forties on no news would move the expected value materially positive without changing a single fact in this document. Third, separation of the two readouts. If the focal-epilepsy extension data lands clearly before or after the bipolar topline rather than inside the same window, attribution.status would move toward CLEAN, the run-up priority score would roughly double, and a clean read on the bipolar result would become possible.

What to watch.

• Now to 2026-10-01: whether the company names a day for the topline, or schedules a conference call. Either would move readout.precision from MONTH to DAY and lift date_confidence from 60 to 100.
• Now to 2026-10-01: whether the focal-epilepsy open-label extension data is scheduled separately. The American Epilepsy Society annual meeting usually falls in early December, which would be outside this window — but the company has not said it will present there, so this is a thing to watch rather than a thing to assume.
• Any Form 4 showing an open-market purchase by an officer or director. Across 227 press items and 85 Form 4 filings this sweep found none. One would be the first genuine conviction signal on this ticker.
• The 2026-11-20 $40 put line. 1,500 contracts of open interest today. Whether that position grows or is closed out ahead of the readout is the one thing in this options chain worth tracking.
• At the topline itself: not the p-value, the size. A placebo-adjusted YMRS difference near the 5-to-7-point range approved drugs achieve (A.3b) is a competitive drug. A statistically significant 2-point difference is a press release, and the market should — and probably will — distinguish the two.
• Immediately after a positive topline: an offering. This company announced its last equity raise on the same day as its last positive topline (../company.md C.3), and both an unlimited automatic shelf and an undrawn $110M at-the-market facility are live.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): miss — probability 35%, band 22–48% [UNVERIFIED — modelled]. The reasoning, since rule 31 requires it: the base rate for a psychiatric Phase 2 proof of concept meeting its primary endpoint sits in the 30–40% region, and acute mania is more tractable than most psychiatric endpoints because approved drugs separate from placebo reliably and by clear margins. Three factors push this specific trial below even. No novel non-dopaminergic mechanism has ever succeeded in acute mania, and the one recent attempt — BHV-7000, 274 patients, 21 days, inpatient, the same YMRS endpoint at the same timepoint — missed outright. Thirteen of the twenty sites running this trial also ran that failed trial, and high placebo response at professional psychiatric research sites is the commonest cause of failure in this indication. And RAP-219 has no human efficacy data in any psychiatric condition; everything supporting it comes from seizure counts in an open-label epilepsy study. Three factors push the other way and are why the band reaches 48% rather than stopping in the thirties: the trial is large and quadruple-masked, so it is not underpowered for a moderate effect; target engagement in living human brain is confirmed rather than assumed; and the sponsor increased enrolment and modified the analysis plan while blinded, which is at least consistent with having seen aggregate data worth investing in. The ~22-day half-life against a 21-day endpoint is a genuine additional risk that no public source resolves. No third-party probability on this event was found.
• Stock-direction (which way do the shares move?): no-edge — confidence low — window 2026-10-01 to 2026-12-15, basis: the readout window (2026-10-01 to 2026-11-30) plus two weeks for the reaction to complete. Materiality is meaningful, not dominant, per ../company.md C.2, and this call is consistent with it — a secondary-indication readout on a company whose dominant value driver reports in 2029 does not warrant a directional bet. attribution.status is CONTAMINATED: the focal-epilepsy open-label extension’s initial data is expected in the same window on the same molecule, so a move inside this window cannot be attributed to the bipolar readout alone.
• Scenario prices: positive $56–$68 · miss $37–$45
• Expected value: $48.35, −1.2% against spot $48.94 (arithmetic, not advice). The no-edge call and this number agree: on this document’s own probability and ranges the event is close to fairly priced, which is why no direction is claimed.
• Run-up: entry $48.94 on 2026-08-19, exit rule T-5 trading days before readout.window.earliest — predicted move −5% to +18%, predicted peak +4% to +30% around 2026-10 — priority score 16, formula_version 1.0.0
• Settles on: the company’s own topline announcement of Phase 2 results from NCT07046494. Positive is defined as: a statistically significant advantage for RAP-219 over placebo (p < 0.05) on the pre-specified primary endpoint, change from baseline in Young Mania Rating Scale total score at Week 3, in the trial’s pre-specified primary analysis population. Anything else — including a numerically favourable result that misses significance, a result significant only on a secondary endpoint or a subgroup, or an unblinded analysis substituted for the pre-specified one — is a miss. Source: Rapport Therapeutics’ own press release or Form 8-K reporting the topline, cross-checked against the ClinicalTrials.gov record for NCT07046494.
• Locked: yes · Settled: no

Program data-quality flags

1. Open Targets search_entities is BLOCKED, four attempts, verbatim Rate limit exceeded for client: global every time — the standing platform throttle 02-connectors.md records. The cost is specific: no independent human genetic evidence on whether CACNG8 or AMPA-receptor genes are implicated in bipolar disorder. A.5’s target-validation row is scored on the sponsor’s own preclinical rationale plus one clinical result in a different disease, and is tagged accordingly.
2. ChEMBL compound_search returned a legitimate empty result for RAP-219 — {"count":0, "total":0,"compounds":[]}, a CALLED state, not a block. Ordinary for an unapproved small molecule whose structure is unpublished, but it means the selectivity claim that the entire mechanism rests on has no independent bioactivity data behind it in this document. Unlike the antibody programs elsewhere in this corpus, where a missing ChEMBL record costs little, this is a small molecule whose whole thesis is regional selectivity — so the gap is more expensive here than the tool’s usual caveat suggests.
3. The dose and regimen are not disclosed by any source read this sweep. The registry says only “RAP-219 tablets administered orally, once daily for 21 days”. Against a ~22-day half-life and a 21-day endpoint this is the single most consequential missing fact in the document (A.1), and it cannot be resolved from anything public.
4. search_investigators returned zero investigators, so A.5b names one person — the trial’s overall official, recovered from the registry record itself — where twenty sites are listed. The conflict search on that one name returned only pre-1990 articles by a different author of the same surname and initials, so conflicts_checked records a search made, never a certification that none exists.
5. No peer-reviewed primary publication of any RAP-219 trial exists. PubMed returned three hits in total; two are conference-summary reviews about epilepsy and one is unrelated. Every efficacy figure in this document comes from a company press release, including the 77.8% seizure reduction and the 24% seizure-free rate that carry A.1’s target-validation argument. That is a materially weaker evidence base than a peer-reviewed report and is the reason A.5 scores publication quality Low.
6. The YMRS benchmark figures in A.3b are [WEB ESTIMATE], not read from primary papers. The asenapine/olanzapine comparison (−10.8 and −12.6 against placebo −5.5 at day 21) and the response rates (42.5%, 52.4%, 29.4%) come from a web search over the published 3-week trial literature rather than from the papers themselves. They set the bar this readout should be judged against, so the limit of that verification is worth stating: the direction and rough magnitude are well established across many trials, but no individual figure here was confirmed against its own source document this sweep.
7. No patent or exclusivity information was located. The dedicated exclusivity search returned the Tenacia licence terms in full and nothing on composition-of-matter claims or expiry dates. B.1’s IP row and B.2’s exclusivity row both record UNKNOWN rather than a guess.
8. The site-overlap finding is verified; the causal reading of it is not. That 13 of 20 sites also ran NCT06419582 is read directly off two ClinicalTrials.gov location lists. That this raises the placebo-response risk is the analyst’s own inference from the general behaviour of acute-mania trials, and it is tagged as such in A.2 rather than presented as a finding.
9. BPIQ row 18870’s catalyst_date_text reads “Q4 2026” while the event behind it is the open-label extension’s initial data. The row’s own indications_text still says “Refractory focal epilepsy (drug-resistant focal onset seizures)” and its stage_event_label still says “Phase 2 Data readout”, both of which described the completed Phase 2a until that trial read out in September 2025. The identification of what the row now covers rests on the company’s 2026-08-05 release naming the extension data for the fourth quarter, not on the row itself. The Attribution computation is unaffected — it uses the row’s date, whatever the row is about — but a reader comparing the pipeline table against the trial list should know the row’s label lags its content.
10. No committed price cache exists for this ticker, so the run-up exit is null, the priced_in driver reads a hand-computed 52-week position rather than one derived from daily bars, and the spot in this document comes from BPIQ rather than from a cache. ../company.md C.4 and C.8 row 10 carry the same limitation at the company tier.