Work
MRTX-1133
KRAS G12D · Reversible, non-covalent, dual-state KRAS G12D inhibitor
Phase 1 of a Phase 1/2 multiple-expansion-cohort trial (63 participants treated). ClinicalTrials.gov records it as Phase 1, terminated 10 Mar 2025; the expansion cohorts never opened. · Terminated / discontinued (public)
- Sponsor / originator
- Mirati Therapeutics
- Current control (public)
- Bristol Myers Squibb (via the 2023 Mirati acquisition)
- Indication
- KRAS G12D solid tumors — pancreatic ductal adenocarcinoma first, also colorectal and NSCLC
- Registry
- NCT05737706
2 material developments may affect this asset.
Why is this asset interesting?
BMS inherited the asset in a $4.8B Mirati deal and discontinued it after Phase 1 on PK, not on a scientific kill. Large-cap portfolio discards of technically failed, biologically intact oncology assets are the classic out-license. The counterparty already said the molecule was not unsafe. What a licensee buys is the right to fix exposure — formulation, prodrug, parenteral route, or a combination that lowers the required dose — and then run the Phase 2 that never started.
- Inferred
KRAS G12D solid tumors — pancreatic ductal adenocarcinoma first, also colorectal and NSCLC
- Verified
Reversible, non-covalent, dual-state KRAS G12D inhibitor
Case file: Mechanism of action
- Inferred
Never approved. First-wave selective G12D small molecule; G12D lacks the cysteine that made sotorasib/adagrasib covalent. A license is to a new product, not a new use of a marketed KRAS drug.
- Inferred
No efficacy verdict exists: NCT05737706 posted no results, and the PDAC, CRC and NSCLC expansion cohorts in its Phase 1/2 design never enrolled. Competitors (RMC-9805/zoldonrasib, degraders, improved non-covalent inhibitors) validate the target while leaving room if this chemotype's delivery is solved. Non-obvious angles: BET-inhibitor combinations reported preclinically, intermittent dosing, and rare G12D tumors outside the three big histologies.
- Inferred
Resurrect surfaced this class because the public record does not look like a conventional drug failure.
| Dimension | Band | Why |
|---|---|---|
| Clinical importance | High | Indication class is life-threatening or high-mortality. |
| Biological rationale | Moderate | Reversible, non-covalent, dual-state KRAS G12D inhibitor |
| Human signal | Unknown | Available trial/stop language does not document a clinical response. Absence of a public signal is not evidence of absence. |
| Safety window | High | Stop language does not name a safety-driven halt, or explicitly disclaims one. |
| Failure recoverability | High | Potentially recoverable |
| Novelty | High | Never reached a US label on the information in this file. |
| Development leverage | Moderate | Highest public stage recorded: Phase 1 of a Phase 1/2 multiple-expansion-cohort trial (63 participants treated). ClinicalTrials.gov records it as Phase 1, terminated 10 Mar 2025; the expansion cohorts never opened.. |
| IP potential | High | Formulation, salt, prodrug or delivery hypotheses are on-thesis for a PK/CMC stop. |
| Dealability | Moderate | Large-pharma control — possible if the program is non-core. Potential licensing candidate; availability is not verified. |
| Evidence quality | Moderate | 4 public source(s) attached to this file. |
Bands are ranking aids, not probabilities of technical or commercial success.
What happened?
Pre-2023
Discovery of a reversible dual-state KRAS G12D inhibitor
InferredMirati developed MRTX-1133 as a non-covalent G12D inhibitor — G12D lacks the cysteine that made sotorasib/adagrasib covalent.
Case file: Reversible, non-covalent, dual-state KRAS G12D inhibitor
2023-02
NCT05737706 posted — Phase 1 of a planned Phase 1/2
VerifiedSolid tumors with KRAS G12D. Phase 2 cohorts were written into the protocol.
ClinicalTrials.gov: NCT05737706
2023
Mirati acquired by Bristol Myers Squibb
InferredPublic deal value cited in secondary reporting at $4.8B. The asset moved with the pipeline. Succession of ownership is inferred from the acquisition, not from a molecule-level assignment document in this file.
Case file: BMS via the 2023 Mirati acquisition
2023-03
Clinical dosing begins
Verified63 participants enrolled in Phase 1. Phase 2 never opened.
ClinicalTrials.gov: NCT05737706 — 63 participants
2025-03
Development discontinued
VerifiedTerminated prior to Phase 2. Secondary reports of a BMS statement: highly variable, less than ideal PK; no safety concerns.
BMS statement (secondary report): Highly variable PK; no safety concerns; discontinued March 2025
Current
Control sits with BMS
InferredNo public evidence in this file that the molecule has been out-licensed. Potential licensing candidate — availability is not verified.
Why did it stop?
Verified facts
- Verified
MRTX-1133 has never been approved.
- Verified
Originator: Mirati Therapeutics. Current control (public): Bristol Myers Squibb (via the 2023 Mirati acquisition). Ownership requires confirmation.
- Verified
Phase 1 of a Phase 1/2 multiple-expansion-cohort trial (63 participants treated). ClinicalTrials.gov records it as Phase 1, terminated 10 Mar 2025; the expansion cohorts never opened.
ClinicalTrials.gov: NCT05737706
- Verified
Technical. The registry's stated reason is “Formulation challenges” (NCT05737706). BMS described the pharmacokinetics as highly variable and suboptimal and reported no safety concerns (secondary reporting). The G12D hypothesis was never tested at a reliable therapeutic exposure.
- Verified
NCT05737706 — “A Phase 1/2 Multiple Expansion Cohort Trial of MRTX1133”; terminated 10 Mar 2025; whyStopped “Formulation challenges”; 63 enrolled; no results posted
- Verified
Bristol exits KRAS G12D — BMS: pharmacokinetics highly variable and suboptimal; no safety concerns identified
- Verified
BMS removes MRTX1133 from its clinical pipeline
Fierce Biotech (secondary report): BMS removes MRTX1133 from its clinical pipeline
- Verified
Prodrug of MRTX1133 as an oral therapy for KRAS G12D — rodent bioavailability improved with lipid formulation
Resurrect analysis
- Inferred
BMS inherited the asset in a $4.8B Mirati deal and discontinued it after Phase 1 on PK, not on a scientific kill. Large-cap portfolio discards of technically failed, biologically intact oncology assets are the classic out-license. The counterparty already said the molecule was not unsafe. What a licensee buys is the right to fix exposure — formulation, prodrug, parenteral route, or a combination that lowers the required dose — and then run the Phase 2 that never started.
- Inferred
No efficacy verdict exists: NCT05737706 posted no results, and the PDAC, CRC and NSCLC expansion cohorts in its Phase 1/2 design never enrolled. Competitors (RMC-9805/zoldonrasib, degraders, improved non-covalent inhibitors) validate the target while leaving room if this chemotype's delivery is solved. Non-obvious angles: BET-inhibitor combinations reported preclinically, intermittent dosing, and rare G12D tumors outside the three big histologies.
- Inferred
G12D is crowded at the target level and empty at the 'this chemotype, this exposure problem' level. A licensee is not racing Amgen on G12C. The race is: can a new presentation of MRTX-1133 (or a backup analog) put patients in a therapeutic window before a covalent or degrader G12D reaches a label.
- Inferred
No publicly identified drug-related safety event appears to have preceded discontinuation. That is an interpretation of the public statement, not a CSR.
Critical unknowns
- Unknown
Human PK tables (Cmax, AUC, variability, food effect): NCT05737706 filed no results section. BMS's description of highly variable, suboptimal PK is the only public PK characterisation. Resolution: Make CSR, bioanalytical listings and formulation composition a condition of any option.
- Unknown
Composition-of-matter and formulation patent status post-Mirati: WO2021/041671 and later formulation filings exist; current assignee, term and lapse are not established here. Resolution: Freedom-to-operate and chain-of-title review before term-sheet economics. Not a model output.
- Unknown
Whether any patient reached a therapeutic exposure: Enrollment was 63; dose and exposure-response are not public. Resolution: Exposure-response from the CSR decides if the next study is a new formulation in all-comers G12D or a salvage of a subset who already got adequate PK.
Is the failure recoverable?
Potentially recoverable. A recoverable category is a reason for expert investigation, not a conclusion that the drug is valuable.
Scientific barrierInferredPotentially recoverable
- What was the problem?
- The G12D hypothesis was not tested at a reliable therapeutic exposure. That is an untested hypothesis, not a negative scientific result.
- Still relevant?
- Yes — G12D remains a high-value oncology target. Competitive agents validate the biology, not this chemotype's exposure.
- Has knowledge or technology changed?
- Selective G12D programs (including covalent and degrader approaches) have advanced since this Phase 1. The target is more, not less, interesting.
- Evidence that would resolve it
- Exposure-response from the CSR: did any patient reach a therapeutic window?
- Experiment or diligence step
- If exposures were subtherapeutic in all comers, the next study is a new presentation, not a repeat of NCT05737706.
Safety barrierInferredPotentially recoverable
- What was the problem?
- Public stop language states there were no safety concerns. No DLT-driven halt is identified.
- Still relevant?
- A private CSR could still contain a signal the public statement omitted.
- Has knowledge or technology changed?
- Nothing public since discontinuation that reverses the 'no safety concerns' statement.
- Evidence that would resolve it
- CSR AE listings, DLT table, and any internal safety memo.
- Experiment or diligence step
- Make the safety listings a condition of any option. If a mechanism-linked signal appears, pass.
Technical / PK barrierVerifiedPotentially recoverable
- What was the problem?
- Highly variable, suboptimal oral pharmacokinetics. The scientific question was not answered because exposure was not controlled.
- Still relevant?
- Yes. This is the reason the program stopped.
- Has knowledge or technology changed?
- Lipid / prodrug presentations of this chemotype have been reported in animals. Whether they translate is unknown.
- Evidence that would resolve it
- Human PK tables (Cmax, AUC, CV%, food effect) and the clinical formulation composition.
- Experiment or diligence step
- Bridging PK/PD in a new presentation before any Phase 2 efficacy spend.
Formulation barrierInferredPotentially recoverable
- What was the problem?
- Oral formulation / bioavailability is the stated technical limit.
- Still relevant?
- Yes, unless a new presentation is shown to work in humans.
- Has knowledge or technology changed?
- ASD, prodrug, parenteral and lipid approaches are more mature than when many first-wave KRAS inhibitors were formulated.
- Evidence that would resolve it
- Whether any presentation produces reproducible therapeutic exposure in humans.
- Experiment or diligence step
- Pick one presentation (prodrug vs parenteral vs ASD) and run a PK go/no-go. Do not run three in parallel as the first spend.
CMC barrierUnknownUnknown
- What was the problem?
- Manufacturing of a new presentation is not described publicly.
- Still relevant?
- Unknown. CMC may be straightforward or may be the hidden reason a reformulation was not attempted.
- Has knowledge or technology changed?
- Not established.
- Evidence that would resolve it
- Drug-substance process, current batch status, and formulation composition.
- Experiment or diligence step
- Request the CMC section of the IND as part of the option package.
Clinical-design barrierInferredPotentially recoverable
- What was the problem?
- Phase 2 cohorts in PDAC, CRC, NSCLC and other G12D tumors were written into the protocol and never enrolled.
- Still relevant?
- The unrun Phase 2 is the development path, not a new idea.
- Has knowledge or technology changed?
- Competitor G12D agents may force a narrower, molecularly defined first cohort.
- Evidence that would resolve it
- Whether any Phase 1 patient had adequate exposure; if yes, a subset salvage is possible.
- Experiment or diligence step
- Do not rewrite a first-in-human. File a bridging PK study, then the Phase 2 that never started.
Biomarker barrierHypothesisPotentially recoverable
- What was the problem?
- G12D itself is the selector. Whether additional PD markers (pERK, downstream) were informative is not public.
- Still relevant?
- Molecular selection is stronger now than at first-in-human design.
- Has knowledge or technology changed?
- Plasma KRAS and PD assays are more available than in 2023.
- Evidence that would resolve it
- Any PD from Phase 1; if none, build a PD package into the bridging study.
- Experiment or diligence step
- Require a PD marker in the bridging PK protocol.
Commercial barrierInferredMixed
- What was the problem?
- G12D is crowded at the target level. A me-too with worse exposure has no role.
- Still relevant?
- Yes — differentiation is delivery of this chemotype, not a new KRAS story.
- Has knowledge or technology changed?
- Competitor G12D agents have moved. A licensee is racing exposure, not Amgen on G12C.
- Evidence that would resolve it
- Can a new presentation put patients in a window before a competitor label.
- Experiment or diligence step
- Competitive landscape memo from patent counsel + clinical; not a model output.
IP barrierUnknownUnknown
- What was the problem?
- Composition-of-matter and formulation families exist (including WO2021/041671). Current assignee, term and lapse are not established here.
- Still relevant?
- Yes. Chain of title after the Mirati acquisition is a gating diligence item.
- Has knowledge or technology changed?
- Not established in this file.
- Evidence that would resolve it
- Freedom-to-operate and chain-of-title review. Not a model output.
- Experiment or diligence step
- Patent counsel on composition, formulation, and whether a new presentation creates a filing position.
Organizational barrierInferredPotentially recoverable
- What was the problem?
- BMS inherited the asset in a large acquisition and discontinued it after Phase 1 on PK. Large-cap discards of technically failed, biologically intact oncology assets are a classic out-license pattern — not a verified offer.
- Still relevant?
- Whether BMS will option a discontinued Mirati clinical asset is unknown.
- Has knowledge or technology changed?
- Development was discontinued (public, 2025). That is not the same as a license being on offer.
- Evidence that would resolve it
- BD counterpart for discontinued Mirati clinical assets; any prior out-license.
- Experiment or diligence step
- Contact. Potential licensing candidate — availability is not verified.
Where is the opportunity?
Labeled Supported / Plausible hypothesis / Requires expert validation. Speculative ideas are not stated as fact.
- Verified
Supported — deadly-illness relevance: KRAS G12D is the dominant KRAS allele in pancreatic cancer, a disease with ~90k US patients and dismal survival. Solving delivery on a G12D-selective inhibitor is a survival question, not a lifecycle management question.
- Inferred
Supported — never-approved novelty: Never approved. First-wave selective G12D small molecule; G12D lacks the cysteine that made sotorasib/adagrasib covalent. A license is to a new product, not a new use of a marketed KRAS drug.
- Hypothesis
Plausible hypothesis — No efficacy verdict exists: NCT05737706 posted no results, and the PDAC, CRC and NSCLC expansion cohorts in its Phase 1/2 design never enrolled. Competitors (RMC-9805/zoldonrasib, degraders, improved non-covalent inhibitors) validate the target while leaving room if this chemotype's delivery is solved. Non-obvious angles: BET-inhibitor combinations reported preclinically, intermittent dosing, and rare G12D tumors outside the three big histologies.
- Hypothesis
Requires expert validation — FDA levers are directional, not a regulatory strategy: Orphan Drug Designation for KRAS G12D pancreatic cancer (prevalence well under 200k); Fast Track / Breakthrough Therapy if a bridging PK study plus an early PDAC signal is generated; Accelerated Approval on ORR in a molecularly defined PDAC or tissue-agnostic G12D cohort, with a confirmatory trial running
- Hypothesis
Requires expert validation — any chemical, formulation, or use modification is a potential IP hypothesis, not a patentability opinion.
Can we get it?
Ownership
- Verified
Historical owner: Mirati Therapeutics.
- Inferred
Current control as recorded: Bristol Myers Squibb (via the 2023 Mirati acquisition). Ownership requires confirmation.
- Unknown
No molecule-level assignment document is in this file. Successor control is inferred from corporate events and pipeline pages.
- Inferred
Evidence of current development: discontinued (public). Asset appears dormant, not active.
Licensing intelligence
Potential licensing candidate — not “available for licensing.”
- Hypothesis
Potential licensing candidate. Do not read this as 'available for licensing'.
- Hypothesis
Who to contact: BD / out-licensing for Bristol Myers Squibb (via the 2023 Mirati acquisition), specifically discontinued or stalled clinical assets.
- Unknown
No public evidence in this file that the asset has already been licensed. Absence of a press release is not proof it is unencumbered.
- Hypothesis
Diligence documents to request: CSR, bioanalytical listings, formulation composition, tox, CMC, chain of title.
IP opportunity
Existing patents
- Inferred
Relevant existing patents: WO2021/041671 and later formulation filings are noted in the case file. Current assignee, term and lapse are not established here.
- Hypothesis
This is not patent counsel and not an FTO opinion.
Potential new IP
Potential IP hypothesis — requires patent counsel and scientific validation.
- Hypothesis
Potential IP hypothesis — requires patent counsel and scientific validation.
- Hypothesis
Formulation / ASD / lipid — scientific rationale: oral exposure was the stop. Conflict: later formulation filings may already cover the obvious presentations.
- Hypothesis
Prodrug — rodent bioavailability improvement has been reported for this chemotype; translation is unknown.
- Hypothesis
Alternate route (parenteral) — bypasses oral variability; may be medically acceptable in PDAC if exposure is solved.
- Hypothesis
Biomarker-selected use / combination (e.g. BET inhibitor, intermittent dosing) — plausible, crowded, requires counsel.
What should we do next?
| Priority | Action | Why it matters | Uncertainty resolved | Evidence |
|---|---|---|---|---|
| High | Open NCT05737706 on the desk, pull the CSR request list, and identify the BMS BD counterpart for discontinued Mirati clinical assets. | Without PK tables and a willing licensor, the rest is a literature review. | Counterparty, package existence, and whether the thesis is even optionable. | Owner correspondence and document index |
| Medium | Option term sheet: exclusive option, CSR + formulation + tox package transfer, limited diligence window, credit of option fee against license. | The only scarce object is the data package. Do not negotiate a full license before seeing exposure-response. | Whether the historical limitation is solvable with a finite experiment. | Primary data package / experiment |
| Low | If PK is salvageable: pick one presentation (prodrug vs parenteral vs ASD), file a pre-IND on a bridging PK/PD study in G12D PDAC, and draft the ODD briefing package. | The first FDA conversation is 'same molecule, new exposure', not a new efficacy protocol that repeats Phase 1. | Whether the historical limitation is solvable with a finite experiment. | Primary data package / experiment |
| High | Pass immediately if a mechanism-based safety finding or an already-licensed exclusive third party appears in the first document drop. | Those are falsifiers, not diligence items to admire. | Whether the file is dead before spend. | Safety listings / license grant |
Material developments
2025-03
What changed
BMS discontinued MRTX-1133 after Phase 1, citing highly variable PK and no safety concerns.
Why it matters
The public stop is technical, not a scientific kill. That is the difference between a pass and an option conversation.
Which conclusion may need review
Recoverability of the exposure problem; whether BMS will option a discontinued Mirati asset.
Post-abandonment
What changed
Competitor G12D programs (including covalent and degrader approaches) continued to validate the target.
Why it matters
Target validation is stronger; chemotype differentiation now rests entirely on solving exposure.
Which conclusion may need review
Novelty versus a future competitor label; whether this chemotype still has a medical role.
Advance to diligence
Generates an editable Asset Acquisition & Development Dossier (28 sections) a BD or translational lead can take into a licensing conversation. Saved to Library.