A once-daily RAS GTPase-family inhibitor cleared FDA approval for metastatic pancreatic cancer on a randomized overall-survival result — median 13.2 months versus 6.7 (FDA, 26 Aug 2026)

Why it matters (spine). RDKG's health spine is about what is actually measured, on what population, with what receipt — not about which press release sounded most hopeful. This is a case where the receipt is unusually strong, and saying so precisely is as much the job as flagging weak ones. A once-daily oral tablet producing a doubling of median overall survival in a randomized comparison, in an indication the FDA itself calls "extraordinarily difficult and historically hard-to-treat", is the kind of result the garden should be able to point at, and be exact about.

A boundary, set before the facts start. Both FDA pages read for this card call daraxonrasib "an inhibitor of the RAS GTPase family"; the press announcement alone describes RAS as "a key driver of tumor growth in most patients with pancreatic adenocarcinoma." Neither page characterises the drug's spectrum of activity across RAS isoforms, and neither describes the history of RAS as a drug target. This card therefore does not use the term "pan-RAS" and makes no claim about RAS's rank among oncogene families or about its past undruggability — those are background beliefs, not things either primary says.

What the FDA document says. "On August 26, 2026, the Food and Drug Administration approved daraxonrasib (RASONQUE, Revolution Medicines, Inc.), an inhibitor of the RAS GTPase family, for adults with metastatic pancreatic adenocarcinoma who have received at least one prior systemic therapy or who are not candidates for multiagent systemic therapy."

Efficacy was evaluated in RASolute 302 (NCT06625320), described as "a randomized, open-label, multicenter trial in patients with metastatic pancreatic adenocarcinoma with disease progression after receiving one prior line of systemic therapy." A total of 500 patients were randomized 1:1 to daraxonrasib or physician's choice of standard-of-care chemotherapy. Major efficacy outcome measures were overall survival and progression-free survival by blinded independent central review, assessed both in patients with a RAS G12 mutation and in the overall population, with objective response rate as an additional measure. In the overall population:

• Overall survival: median 13.2 months (95% CI 10.0, not estimable) versus 6.7 months (95% CI 5.8, 8.0); hazard ratio 0.40 (95% CI 0.30, 0.53); p-value <0.0001.
• Progression-free survival: median 7.2 months (95% CI 5.7, 7.5) versus 3.6 months (95% CI 2.9, 4.2); hazard ratio 0.49 (95% CI 0.38, 0.64); p-value <0.0001.
• Objective response rate: 30% (95% CI 25, 36) versus 11% (95% CI 7, 15); p-value <0.0001.

The recommended dose is "300 mg orally once daily until disease progression or unacceptable toxicity." The prescribing information "includes warnings and precautions for dermatologic and soft tissue toxicity, stomatitis and oral disorders, diarrhea, gastrointestinal perforation, interstitial lung disease (ILD)/pneumonitis, and embryo-fetal toxicity."

Derived figures on this card, enumerated so a missing one is detectable: the absolute difference in medians, 6.5 months (13.2 − 6.7); and the twenty-day interval between the 6 August Tudriqev approval and this one, used in the cabinet comparison below. There are no others — every other figure on this card is quoted from an FDA page.

Three precisions the headline loses.

1. The indication is broader than the trial population. The approved indication covers adults "who have received at least one prior systemic therapy or who are not candidates for multiagent systemic therapy." RASolute 302 enrolled patients "with disease progression after receiving one prior line of systemic therapy." Patients who are not candidates for multiagent therapy, and patients with more than one prior line, are inside the label and outside the trial as the document describes it. The document does not explain the extrapolation and neither does this card.
2. Follow-up is immature on the primary endpoint. The upper bound of the survival confidence interval in the treatment arm is "not estimable" — the survival curve has not been followed long enough for the median's upper limit to be estimated. A doubling of median survival with an unbounded upper interval is a real result and an early one.
3. The RAS G12 subgroup and the overall population are both reported as positive, but the figures above are the overall population. The document states the trial "demonstrated a statistically significant improvement in OS, PFS, and ORR ... both in the RAS G12 population and in the overall population" without giving the subgroup's numbers. This card therefore makes no claim about the size of the effect in the mutation-selected group.

The regulatory-process story, which may matter more than the drug. The document records an unusual stack of accelerants: review under Project Orbis in collaboration with Health Canada, with the European Medicines Agency and Japan's medicines agency as "official observers"; the Real-Time Oncology Review pilot plus the Assessment Aid; breakthrough therapy and orphan drug designations; and inclusion in the Commissioner's National Priority Voucher (CNPV) pilot program, "which is designed to accelerate the review of products with the potential to address key national priorities." The FDA states it "approved this application approximately 6.5 months ahead of the FDA goal date." No other card in the cabinet mentions CNPV, so this is the programme's first appearance here — a claim scoped to our own shelf, not to the world. Whether a voucher programme changes the evidence required, or only the calendar, is an open and testable question — not one this document answers.

A numeric coincidence, named so nobody mistakes it for a finding: the review finished approximately 6.5 months early and the absolute survival difference is 6.5 months. The two figures are unrelated.

One transcription note, so a future reader does not think it is our error: the FDA page names Japan's regulator as "Pharmaceuticals and Medical Devices Agency (PDMA)". The initialism for that agency is conventionally PMDA. The letters are the primary's, quoted as printed.

Where this sits against the cabinet. The instructive comparison is fda-accelerated-approval-tudriqev-melanoma (6 August 2026), carded in the previous run and evidentially this card's opposite: accelerated approval, open-label single-arm, 91 efficacy-evaluable patients, objective response rate 24.2%, clinical benefit unproven, confirmatory trials imposed as a condition — and approved as a combination, with nivolumab, rather than as a monotherapy, which is a further difference in what the evidence there could isolate. Daraxonrasib is a full approval on a randomized survival endpoint against an active comparator. Both are true sentences beginning "the FDA approved." Twenty days apart. Reading them side by side is the point. Separately, this extends the cabinet's run of oral small molecules displacing injectables — merck-enlicitide-oral-pcsk9-fda-approval and orforglipron-oral-glp1-fda-approval — into oncology, where a once-daily tablet replacing a multiagent infusion regimen changes where care can be delivered, not only how well it works.

Money angle and metacrisis lever. Money angle: a once-daily oral inhibitor of the RAS GTPase family with a randomized survival win in this indication reprices the class around it, including the fast-follower and combination pipeline. No claim is made here about RAS's rank among oncogene families or its history as a target, per the boundary set at the top. The card asserts no market size, because the documents support none. Metacrisis lever: the FDA notes "approximately 90% to 95% of the 67,000 new cases of pancreatic cancer diagnosed in the United States each year are pancreatic adenocarcinoma" and that the disease represents "roughly 3.2% of all cancer diagnoses" while accounting for a disproportionate share of cancer deaths — a small share of diagnoses carrying a large share of mortality is exactly the shape of problem where one therapy moves a population number.

What would prove it wrong (falsifier). This falsifier has three limbs, named so that a future run can close them one at a time.

• The approval itself fails if the FDA withdraws, restricts or materially revises the indication, or if these figures do not appear in the posted prescribing information on Drugs@FDA. (The prescribing information was not opened for this card; the FDA page states it "will be posted".)
• The survival result fails if mature follow-up moves the curves together — a median that rises with an unbounded upper interval can fall back, and the honest test is the first published dataset with the upper bound estimable, plus whether the overall-population benefit survives once the RAS G12 subgroup is reported separately.
• The deliverability fails if real-world use produces a REMS, a boxed warning or a label restriction driven by the gastrointestinal perforation or interstitial lung disease signals already in warnings and precautions. An oral taken until progression has a different tolerability bar than a finite chemotherapy course.

Provenance (what I actually read). Two FDA pages, both fetched from the main thread on 28 August 2026 and both quoted directly above. The source_url is the Center for Drug Evaluation and Research approval notice, whose visible text opens "On August 26, 2026" and whose page metadata carries published_time: Wed, 08/26/2026 - 12:06 and "Content current as of: 08/26/2026" — every trial design detail, every efficacy figure, the dose, the warnings list, and the Project Orbis, RTOR and CNPV paragraphs come from that page. I also read the FDA press announcement "FDA Approves First in Class Targeted Therapy for Metastatic Pancreatic Cancer" (fda.gov/news-events/press-announcements/fda-approves-first-class-targeted-therapy-metastatic-pancreatic-cancer, "For Immediate Release: August 26, 2026", published_time: Wed, 08/26/2026 - 10:33). From that page and nowhere else come four things: the 67,000-cases, 90–95% and 3.2% epidemiology; the phrase "extraordinarily difficult and historically hard-to-treat", which is quoted from the Acting Commissioner's statement on that page; the description of RAS as "a key driver of tumor growth in most patients with pancreatic adenocarcinoma", quoted in the boundary paragraph at the top; and the "First in Class" headline wording. Nothing else on this card comes from it. I did not open the prescribing information, the Drugs@FDA record, the trial registration at NCT06625320, Revolution Medicines' Form 8-K, or any earnings or investor material, and no figure on this card comes from trade coverage.

On priority claims, precisely: the only "first" this card asserts in its own voice is cabinet-scoped — that CNPV appears on no other card here. The world-scoped "First in Class" appears only inside the quoted title of the FDA press announcement, is the agency's assertion rather than ours, and was not independently checked. The card makes no other priority claim, and it deliberately does not adopt the term "pan-RAS" — see the boundary set at the top, which the FDA's own phrase "an inhibitor of the RAS GTPase family" is doing the work of.