TYRA / dabogratinib-ir-nmibc — Dabogratinib (TYRA-300) for low-grade, intermediate-risk non-muscle invasive bladder cancer

Program analysis · bpiq_drug_id 18677 · prepared 2026-08 · USD · framework v5.5.0 · 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): An oral, once-daily pill designed to block only FGFR3 — the growth switch mutated in ~70% of intermediate-risk NMIBC — attempting to dissolve bladder tumors that today are removed surgically, over and over.
• The event and when (as disclosed): Initial SURF302 data — safety on >40 patients, efficacy (3-month complete response) on >20 — guided to “September 2026” [VERIFIED — Q2 2026 release, 2026-08-04]. The guidance has slipped twice, from 1H 2026 to August to September.
• The main reason it could work: The biology is validated end to end: the same drug produced a 54.5% response rate in far harder-to-treat metastatic disease, an FGFR-targeted competitor (TAR-210, locally delivered erdafitinib) produced ~90% CR in this exact population, and low-grade papillary tumors are the most FGFR3-dependent tumors known.
• The main risk: Expectations, not biology. The approved bar is 78% CR; the FGFR-selected bar is ~90%; the stock has re-rated 2.7× on anticipation; and the only efficacy disclosure in this company’s history — data management called positive — dropped the stock 23% in a day. A CR rate in the 50s or 60s is simultaneously a scientific success and a probable stock miss.
• What it means for the stock: No edge on direction. The expected value sits ~6% above spot and the sign flips inside the probability band. The window is also contaminated: TYRA-200’s Q4 2026 readout can land inside it, so a late move is not attributable to SURF302 alone.

0. Program-tier coverage — CLEARED

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

A. Scientific & clinical assessment

A.1 Mechanism of action and scientific rationale

Dabogratinib is a small-molecule pill taken once a day. It blocks FGFR3 — fibroblast growth factor receptor 3 — a protein switch on the surface of bladder-lining cells. In roughly 70% of intermediate-risk NMIBC the FGFR3 gene carries a mutation (most often S249C, or Y373C — together about 80% of the mutations seen) or a fusion that welds the switch permanently on, driving the constant regrowth of low-grade papillary tumors [VERIFIED — Mol Cancer Ther 2026, DOI 10.1158/1535-7163.MCT-25-0652].

The design point is what dabogratinib does not hit. Older “pan-FGFR” drugs such as erdafitinib block FGFR1, 2, 3 and 4 together, and the off-target hits cause the side effects that limit them: blocking FGFR1 in the kidney raises blood phosphate in more than 70% of erdafitinib patients, and mouth sores, nail damage, eye problems and diarrhea led to dose interruptions in 68% and permanent discontinuation in 21% in metastatic use [VERIFIED — same paper, citing the erdafitinib label trial]. Dabogratinib is 63-fold selective for FGFR3 over FGFR1, 19-fold over FGFR2 and 55-fold over FGFR4 in cellular assays, and was built to keep working against the V555M “gatekeeper” resistance mutation that defeats the pan-FGFR class [VERIFIED — J Med Chem 2024, DOI 10.1021/acs.jmedchem.4c01531 — the company's own discovery paper; no independent replication exists, see data-quality flags].

The step the September readout must prove is chemoablation: that swallowing this pill makes visible low-grade bladder tumors disappear, measured as complete response at 3 months on a marker lesion deliberately left after surgery. Evidence it might: in the SURF301 trial the same molecule produced 6 confirmed partial responses in 11 evaluable patients (54.5%) with metastatic FGFR3-altered urothelial carcinoma at doses ≥90 mg — a much harder setting than a superficial low-grade tumor — with responses lasting 7.4 to 12.6 months in the published cases and no treatment-related grade ≥3 events in those three patients [VERIFIED — DOI 10.1158/1535-7163.MCT-25-0652]. And J&J’s TAR-210, which bathes the bladder in erdafitinib from an implanted device, produced ~90% CR in FGFR-altered NMIBC — direct proof that FGFR inhibition ablates these tumors [WEB ESTIMATE — UroToday/OncLive conference coverage, read 2026-08-07].

The honest scientific risk: SURF302 uses lower doses (50 and 60 mg) than the ≥90 mg that produced the mUC responses, betting that superficial disease needs less drug; systemic oral exposure must reach the bladder lining at tumor-killing levels without the local concentrations an intravesical device achieves; and open-label small-n CR rates are noisy. None of these is a mechanism doubt. All of them are bar-clearing doubts.

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

Trials

Feasibility matrix

Resourcing sufficiency. Not a constraint. Cash of $353.9M runs into 2H 2028, roughly two years past this catalyst, all three Phase 2 trials are funded, and R&D spend is scaling (+61% year over year) rather than being rationed. See ../company.md C.3 [VERIFIED — Q2 2026 release].

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

A.3a Target label. Oral treatment of FGFR3-altered low-grade, intermediate-risk non-muscle invasive bladder cancer in adults — a biomarker-defined slice (~70%) of the recurrent LG-IR-NMIBC population.

A.3b Target product profile

A.3c Strategic Go/No-Go questions. (Pre-Phase-III set — the next decision is whether SURF302 supports a registration path.)

A.3d Regulatory designations. None found for dabogratinib in NMIBC — no Fast Track, Breakthrough or Orphan designation appears in the 236-item press feed or the trial record. (The achondroplasia program’s designations, if any, belong to that program, not this one.) [UNVERIFIED — absence asserted from the swept sources only]

A.4 Target product profile valuation matrix

A.5 Evidence quality and endpoints

Endpoints

A.5b Key opinion leaders.

CT.gov names no overall official and no investigator on any of SURF302’s 47 locations, so the investigator panel below is built from the program’s peer-reviewed clinical publication (SURF301), whose named academic authors are trial investigators at their sites. No independent voice on this program’s endpoint was verified this sweep — that is a recorded absence, not an unexamined one; the searches are listed per row and in the program data-quality flags.

Panel as of. 2026-08-07.

Investigators

Independent voices

No independent voice is recorded. The searches run (PubMed on the program’s five records; web search of conference coverage) surfaced named commentators only in competitor-trial contexts — e.g., TAR-210 investigators — who are not independent of the competing sponsor, and recording a conflicted voice as independent is the contradiction the schema forbids.

Judgement

Dissent

B. Commercial assessment

B.0 Current treatment algorithm

1. A bladder tumor is found (usually after blood in the urine) and removed by TURBT — surgery through the urethra under anesthesia. Pathology grades it: low-grade Ta/T1, intermediate risk per AUA 2024 criteria if it recurred within a year, is >3 cm, or is multifocal.
2. Today the LG-IR patient then enters surveillance: cystoscopy every few months, and another TURBT every time tumors regrow — which for intermediate-risk disease is the expected course. A single dose of intravesical chemotherapy after surgery reduces early recurrence; BCG (the immune-stimulating bladder infusion) is standard for high-risk, not this population.
3. Since 2025-06-12 there is one approved drug alternative to repeat surgery: UGN-102 (Zusduri), a mitomycin gel instilled through a catheter in six weekly office visits — 78% CR at 3 months [VERIFIED — FDA].
4. Where dabogratinib would fit: in the ~70% of these patients whose tumors carry an FGFR3 alteration, as the first at-home oral option — displacing repeat TURBT and competing directly with UGN-102 and, if approved, TAR-210 for the “treat instead of operate” slot. It adds a biomarker-testing step that neither surgery nor UGN-102 requires.

B.1 Competitive landscape and positioning

Where this asset wins: the ~70% FGFR3-altered slice of a population that hates its current treatment, with the only option that requires neither operating room nor catheter. The single fact the thesis rests on: oral, systemic dabogratinib at 50–60 mg must ablate superficial tumors roughly as well as drugs delivered directly into the bladder. If September’s CR rate is near the intravesical bar, the convenience story writes itself; if it is materially below, “oral” stops being an advantage and becomes the explanation.

B.2 Addressable market

B.3 Value and feasibility

B.3a Expected peak sales. Built bottom-up as a labelled model; no verified epidemiology or pricing input was obtained this sweep, so every scenario is a model, conditional on approval in IR NMIBC (roughly 2029+ given primary completion 2028-02), US-only, this indication only — excluding UTUC, achondroplasia, and ex-US.

The inputs that are not defensible today, named per rule 30/9: the treated-population size (no verified epidemiology of the recurrent, tested, FGFR3+ slice), the price (no verified Zusduri course price obtained, no dabogratinib pricing signal), and TAR-210’s share split. Sell-side raised NMIBC peak-sales models in 2026 without publishing figures [WEB ESTIMATE — Investing.com Oppenheimer note, read 2026-08-07].

B.3b Feasibility

B.4 Product-development risk

Framing questions.

1. Does the development plan support the target product profile claims? Mostly. The open-label marker-lesion Phase 2 directly measures the claim that matters (chemoablation CR). What it cannot support is the comparative convenience-at-parity claim — that needs the CR number to land near the intravesical bar.
2. Will the identified risks affect the target product profile? The dose-selection risk does: if 50/60 mg underperforms because ≥90 mg was the active dose, the profile’s tolerability claim collides with its efficacy claim, since higher doses brought dose reductions in SURF301 cases.
3. If a risk cannot be mitigated, is the asset still differentiated? Yes in one direction only: even a middling-CR oral remains the only at-home option — but payers and urologists have two higher-CR local options, so differentiation without competitive efficacy is a niche, not a franchise.

Readout

Sources

The modelled estimate. This catalyst is an interim the company controls, so there is no registry anchor to add a lag to; the arithmetic is enrollment-driven instead. More than 40 patients were enrolled by 2026-08-04 with first patient dosed 2025-06-27; every patient dosed by June 2026 has a 3-month CR assessment by September 2026, so an efficacy-evaluable set of >20 is mechanically available in September, and a cut-plus-cleaning allowance of zero to two months puts the plausible disclosure range at September to November 2026. data/benchmarks/readout-lag.json holds no comparable interim-disclosure observation, and the default primary-completion lag of rule 34 does not apply to a company-controlled interim — stated rather than misapplied.

Window

Basis. The company names a month, twice re-guided, so the window opens on that month’s first day. Likeliest mid-September: the September options positioning (C.6) and the “September marks an important milestone” framing in the Q2 release read as a date the company intends to hit, and an interim it controls needs no long lag. Latest end-October: each of the two prior slips was exactly one month (1H 2026 → August → September), so one more identical slip is retained as the conservative edge. Precision is MONTH because “September 2026” names a month and no source names a day. Confidence is MEDIUM, not HIGH, because the same guidance has already moved twice in two quarters.

Disagreement. CONSISTENT. bpiq mirrors the company’s wording; the modelled range contains it; ctgov dates a different milestone (full primary completion, 2028-02) and therefore neither confirms nor contradicts the interim date.

Date slippage. Two slips in four dated statements.

Attribution

Computed by lib/clustering.mjs attributionFor over every has_catalyst row in ../company.md C.2, with this program’s own readout window, at CATALYST_CLUSTER_MIN_MONTHS = 6; output transcribed.

Status. CONTAMINATED

Conflicts

Note. Both conflicts are confirmed through this program’s own MONTH-precision window — the other side of each pair is a period placeholder. The one that bites is TYRA-200 (17139): its readout is guided to Q4 2026 in the company’s own text, so it can land as early as October, inside this window’s back half and squarely inside the stock call’s reaction window; a move in October–November is not attributable to SURF302 alone. BEACH301 (16756) is guided to end of Q1 2027 and its half-year placeholder merely sits 62 days past this window’s latest edge — it contaminates the arithmetic, not really the September story. The scenario ranges below are therefore written for the September–October front half of the window, where SURF302 is the only plausible mover on this ticker.

Market and timing for this event

• Plain takeaway. A heavily pre-traded, heavily shorted, crowd-owned binary landing inside eight weeks, priced at 55% of a 52-week range it climbed on anticipation alone, with the only historical efficacy print on this ticker having resolved −23% in a day.
• Months to this catalyst. 0.8 to readout.window.earliest (2026-09-01) and 2.8 to the latest edge (2026-10-31), from 2026-08-07. Precision is MONTH, so the true gap is known to within a few weeks, not days.
• Expected move around this event. The chain cannot price it cleanly (../company.md C.6): the September $25 straddle brackets 17–49% of spot bid-to-ask. Stated as a bracket: a 25–40% move on data is what the chain’s wide quotes and the ticker’s own history jointly suggest; no point estimate is defensible.
• Nearest comparable past reaction. C.7’s 2024-10-25 row — the SURF301 interim (54.5% PR in mUC), the only efficacy disclosure in company history: −23.3% close-to-close, −40.9% over three sessions, from a similar anticipation-inflated base. It is the direct analogue: same drug, same sponsor, “positive” headline, expectations unmet. The positive analogues (+14.3%, +5.2%) are dosing milestones and do not scale to an efficacy print.
• Materiality. Dominant near-term per ../company.md C.2 — the next disclosed catalyst on the whole pipeline, though not the whole equity. Attribution is CONTAMINATED (above): the call below is presented as attributable to SURF302 only in the window’s front half; the back half is shared with TYRA-200’s guided Q4 window.
• Date slippage. 2 slips (Readout, above) — both one month, both within 2026.

Spot. $26.96, read 2026-08-07 (BPIQ last price, equal to the price cache’s 2026-08-06 close), from ../company.md C.4.

Scenario prices

Expected value. 0.55 × $38.00 (positive midpoint) + 0.45 × $17.00 (miss midpoint) = $28.55, +5.9% against spot $26.96. Method: probability_pct applied to the midpoint of each scenario range. At the band edges the sign flips: $25.40 (−5.8%) at 40%, $31.28 (+16.0%) at 68%. This is arithmetic, not advice, and it is not a price target.

Run-up

Entry and exit

Predicted move and peak

Priority score drivers

Five of the seven transcribed from lib/runup.mjs scoreDriver; the two judgement drivers read from this document.

Priority score. From lib/runup.mjs priorityScore over the seven rows above.

Priority score 21 · formula_version 1.0.0

Settlement. Left null at lock time. The exit could not be resolved either: the committed price cache ends 2026-08-06, before the window opens, so resolveExit returns null and exit is recorded as null on the prediction.

Verdict

What I would do. Watch. Do not chase the run-up; reassess on the print.

Why. The science is the good part: the target is the best-validated driver in this tumor type, the same molecule already ablated harder disease, and the endpoint is an approved registration currency. The trade is the bad part. The expected value sits only ~6% above spot and flips sign inside its own probability band, because the market has spent fourteen months and 2.7× paying for this readout in advance while the bar was being raised externally — 78% CR approved (UGN-102), ~90% in early FGFR-selected data (TAR-210). The one time this company printed efficacy data into inflated expectations, the stock fell 23% in a day on a result management called positive. And a win does not pay cleanly: a $250M ATM was refreshed one month before the readout, and the back half of the reaction window is shared with TYRA-200’s Q4 readout.

What would change this. Upward: the initial CR rate landing ≥70% with single-digit grade ≥3 toxicity — that converts the convenience thesis into a competitive-efficacy thesis and the 32% short position into fuel; buy the confirmation, not the anticipation. Downward before the print: a third guidance slip past October, which would break the “named month” premise this window and run-up call rest on.

What to watch.

• 2026-09-01 to 2026-09-30 — the guided disclosure itself: CR rate, evaluable n per dose, discontinuations at 50/60 mg, any phosphate or ocular signal.
• Any date — an ATM drawdown or offering filing before the data (would signal management selling ahead of, not into, the print).
• End of 2026 — MoonRISe-1 (TAR-210) enrollment completion; its Phase 3 cadence defines the competitive clock this program races.
• Q4 2026 — TYRA-200 ICC initial data (bpiq 17139): the attribution contaminant; a move on this ticker after October needs decomposing before it is read as SURF302’s.
• 2028-02 — SURF302 registry primary completion; if it moves, the program’s full-data cadence moves with it.

Locked prediction

• Outcome-direction (will the readout succeed on its primary endpoint?): positive — probability 55%, band 40–68% [UNVERIFIED — modelled] — the probability that the settlement definition below (initial CR rate ≥60%, no stopping safety signal) is met.
• Stock-direction (which way do the shares move?): no-edge — confidence low — window 2026-08-07 to 2026-11-30, basis: the readout window (earliest 2026-09-01, latest 2026-10-31, MONTH precision) plus four weeks for the reaction to settle; expected value +5.9% with a sign-flip inside the probability band; attribution CONTAMINATED in the window’s back half.
• Scenario prices: positive $30.00–$46.00 · miss $14.00–$20.00 (from the table above)
• Expected value: $28.55, +5.9% against spot $26.96 (arithmetic, not advice)
• Run-up: entry $26.96 on 2026-08-07, exit rule T-5 trading days before readout.window.earliest — predicted move 0–10%, predicted peak 10–30% around 2026-09 — priority score 21, formula_version 1.0.0
• Settles on: the first public disclosure of initial SURF302 (NCT06995677) results, guided September 2026; readout window 2026-09-01 to 2026-10-31. Positive definition: the disclosed 3-month complete response rate in efficacy-evaluable FGFR3-altered patients, aggregated across the 50 mg and 60 mg once-daily arms, is ≥60%, and no treatment-related death, dosing hold or program discontinuation in IR NMIBC is announced with it. Anything else is a miss — including an aggregate CR below 60%, or data disclosed without an evaluable CR rate. Source: Tyra’s own press release or presentation, cross-checked against the ClinicalTrials.gov record.
• Locked: yes · Settled: no

Program data-quality flags

• ChEMBL holds no record of dabogratinib under “dabogratinib”, “TYRA-300” or “TYRA300” (three calls, all legitimately empty). Molecular identity and the 63×/19×/55× selectivity ratios rest entirely on Tyra’s own peer-reviewed discovery paper (DOI 10.1021/acs.jmedchem.4c01531); no independent replication of selectivity exists in any swept source.
• Open Targets rate-limited on first attempt (verbatim: Rate limit exceeded for client: global); the immediate retry succeeded, so the state is CALLED. “Non-muscle invasive bladder cancer” returns no disease entity — target-level validation only.
• CT.gov names no investigator or contact on any of SURF302’s 47 locations. The KOL investigator panel is built from the SURF301 publication’s named academic authors instead, with each person’s verifiable trial attachment stated per row.
• PMC full text strips back matter: the dabogratinib clinical paper (PMC13060625) came back without its Declarations section, so its own COI statement was not retrievable through this connector. Recorded in each conflicts_checked entry rather than papered over.
• This catalyst is a company-controlled interim, not a registered milestone. The registry’s primary completion (2028-02) does not date it, and no independent source can. The readout window rests on the company’s twice-slipped guidance plus enrollment arithmetic — stated in the Readout section rather than presented as externally anchored.
• The BPIQ note field carries only one dated entry for this program; the slip sequence was reconstructed from the company’s own release wording via web search of the ir.tyra.bio/ PRNewswire record and is quoted with its dates in Readout.
• Registry vs release dose discrepancy, minor: NCT06995677 lists a third “Dose TBD” arm; the Q2 2026 release describes “two QD doses (50 mg and 60 mg)” and guides initial data on both. Read as an unopened escalation arm; initial data cover the two QD doses.
• No verified epidemiology or pricing input for the B.3a model could be obtained this sweep; all three peak-sales scenarios are labelled models, and the missing inputs are named there.
• Competitor figures are taken as published headlines (ENVISION 78%/79%; TAR-210 ~90%/89%; erdafitinib toxicity rates). The ENVISION and erdafitinib figures trace to FDA/paper sources; the TAR-210 figures are conference coverage and are tagged WEB ESTIMATE wherever used.