The paper appeared on May 6. Brian M. Wolpin, of the Hale Family Center for Pancreatic Cancer Research at Dana-Farber, was first author. Twenty more investigators followed. The journal was the New England Journal of Medicine. The DOI is 10.1056/NEJMoa2505783. The phase 1–2 cohort enrolled 168 patients with previously treated, RAS-mutated, metastatic pancreatic ductal adenocarcinoma. The 300-milligram once-daily dose produced, in the 26-patient second-line G12 subgroup, an objective response rate of 35 per cent and a median overall survival of 13.1 months. [1] The paper's May 13 account of the phase 3 RASolute 302 topline — 13.2 versus 6.7 months median overall survival — has, eight days later, a peer-reviewed underpinning.

It also has a peer-reviewed critic. Two days earlier, on May 4, Timothée Olivier — Vinay Prasad's regular co-author on questions of cancer-drug response-rate denominators — published a piece in The Oncology Shot tightening the denominator. Olivier's argument is that the headline ORR figure quoted in press accounts (20 to 27 per cent across various reads of the dataset) inflates the response by counting unconfirmed responses, mixing dose levels, and folding in radiographic improvements that did not meet the threshold for a confirmed objective response per RECIST 1.1. When Olivier restricted the count to confirmed responses among patients receiving the recommended phase 3 dose, the ORR fell to 13.4 per cent. [2] Either number is a real number. They measure different things.

This is unusual. A drug arrives at ASCO's Plenary on May 31 with both the publication and the published response-rate critique already on the table. The plenary discussant — yet unnamed — will have to address Olivier's denominator math because the analysis is on the record at the same time the data are.

The framing matters because daraxonrasib's full story is not yet a confirmed approval. The May 1 FDA "safe to proceed" letter authorising expanded access (covered in the paper's parallel May 13 piece) is not an approval. The phase 3 RASolute 302 readout on April 13 is unprecedented for metastatic PDAC but is also a second-line readout. The first-line trial — RASolute 303 — is still enrolling, with a three-arm design that places daraxonrasib monotherapy and daraxonrasib-plus-chemotherapy against gemcitabine plus nab-paclitaxel. [3] The first-line story, in other words, is not yet decided. The second-line gap — 13.2 versus 6.7 months — has been moved by a single trial. It has not been moved twice.

What Olivier's piece actually says, when read carefully, is not that daraxonrasib does not work. It says that the framing of "miracle drug" depends on which slice of the phase 1–2 cohort one uses. The 35 per cent ORR in the second-line G12 subgroup is the number in the NEJM abstract. The 29 per cent ORR in the 38-patient any-RAS-mutation second-line cohort, also at 300 milligrams, is the broader-population number. The published NEJM data give both. [1] Olivier's 13.4 per cent confirmed-ORR number derives from a different slice — including dose levels below 300 milligrams and counting only patients with formally confirmed responses on follow-up imaging. Treatment-related grade 3 or higher adverse events were reported in 30 per cent of the 168-patient cohort, which is the toxicity flag the ASCO Post highlighted in interviews with both Wolpin and his Dana-Farber colleague Colin Weekes. [4]

The Wolpin paper itself frames the data with restraint. The conclusions section states that daraxonrasib "was associated with treatment-related adverse events of grade 3 or higher in one third of patients" and that "antitumor activity was also reported." [1] The authors explicitly note that direct comparisons cannot be made to standard chemotherapy because the phase 1–2 cohort lacks a control arm. The randomised second-line comparison is what RASolute 302 provided in April. The phase 1–2 paper, in other words, is the platform that justified the phase 3 trial. The phase 3 trial is the evidence base for the survival claim.

The drug's mechanism explains some of the response-rate discrepancy between papers and presentations. Daraxonrasib is a RAS(ON) multi-selective inhibitor — it binds active, GTP-bound mutant and wild-type RAS, blocking interactions with downstream effectors. RAS mutations are present in approximately 90 per cent of pancreatic ductal adenocarcinoma cases. The drug's design is genuinely novel; the mechanism is genuinely on. The question is the durability of response and the comparability of responder-fraction definitions across trial designs. That question requires the full RASolute 302 dataset, which will be presented at ASCO Plenary on May 31 by Wolpin himself, who is on the steering committee for that trial. [5]

There is also a denominator-of-the-denominator question that the paper, the topline, and Olivier all share: the population is exclusively patients who progressed on prior treatment. RASolute 303, the first-line trial that began enrolling in Q4 2025, will measure the drug against the standard of care in patients who have never received chemotherapy. The presentation at the 2026 AACR Annual Meeting on April 21 showed a 47 per cent ORR in the first-line monotherapy cohort and a 58 per cent ORR in the first-line combination cohort — but those are small phase 1–2 numbers (38 and 40 patients respectively), and the six-month progression-free survival rate was 71 to 84 per cent. [6] Encouraging. Not yet randomised.

Revolution Medicines' own framing has shifted. The May 6 press release accompanying the NEJM publication used language carefully: "promising clinical antitumor activity and durable responses, with an acceptable safety and tolerability profile, in patients with previously treated metastatic RAS mutant PDAC." [5] Note the qualifiers — "promising," "acceptable," "previously treated." The CEO Mark Goldsmith's April 13 phase 3 quote, "potentially transformative advance," is also on the record from a month earlier. The framing escalates with the data. The data have escalated. So has the critique.

The ASCO Plenary is 17 days away. The plenary slot is one of the highest-visibility positions in oncology and the place where Wolpin will present the full RASolute 302 dataset including subgroup analyses, depth of response, duration of response curves and quality-of-life data. The plenary discussant — convention dictates a senior oncologist not affiliated with the trial or sponsor — will deliver a commentary that the field will read as the authoritative interpretation. Whether that commentary names Olivier's denominator math, and whether it agrees or disagrees, will shape how the FDA reviews the eventual NDA submission. The FDA's Commissioner's National Priority Voucher pathway, which Revolution has been granted, accelerates review timelines but does not change the analytical standards.

The deeper test is whether oncology can hold two true things at once — that daraxonrasib has, in a single phase 3 trial, doubled median overall survival in the most resistant cancer in adult oncology; and that the response-rate framing in some press accounts has been inflated relative to the confirmed-response math. Both are true. Neither cancels the other. The drug is real. The math is real. The Plenary will tell us whether the field is ready to litigate them together.

-- KENJI NAKAMURA, Tokyo