On August 26, 2026, the US Food and Drug Administration approved daraxonrasib for adults with metastatic pancreatic cancer that had already been treated once [1]. On its own, one approval is a headline that fades within a week. Read alongside four other developments from the same few years, it looks more like a change in how quickly cancer research now reaches patients.
This guide covers the drug, the protein it targets, the trial behind the approval, the speed of the regulatory process, and the personalized cancer vaccine result announced the week before. Every figure is drawn from the published papers and press releases listed at the end.
Table of Contents
- What is daraxonrasib and what did the FDA approve?
- Why was RAS called undruggable for forty years?
- How does daraxonrasib work differently from earlier RAS drugs?
- What did the RASolute 302 trial show?
- How fast was the approval, and who could get the drug before it?
- Which other cancers carry RAS mutations?
- Where did cancer immunotherapy begin?
- Why did earlier cancer vaccines fail?
- How is a personalized cancer vaccine made?
- What did the INTerpath-001 melanoma trial find?
- Why did cancer research have a reproducibility problem?
- What do these pieces add up to?
- References
What is daraxonrasib and what did the FDA approve?
Daraxonrasib is an oral tablet made by Revolution Medicines. It blocks a family of proteins called RAS, which act as the growth switch inside cells. The FDA approval covers adults whose pancreatic adenocarcinoma has spread and who have had at least one prior round of systemic treatment, or who cannot tolerate combination chemotherapy [1].
The approval rested on a single phase 3 trial, RASolute 302, which compared the tablet with chemotherapy in 500 patients. The result is described in detail below. What sets this approval apart is less the drug itself than the protein it blocks, which had defeated every attempt at a drug for four decades.
Why was RAS called undruggable for forty years?
The story starts with a bladder tumor. In 1982, researchers found that a single-letter change in the DNA of a human bladder cancer was enough to convert a normal gene into one that drives uncontrolled growth [4]. The gene was a member of the RAS family.
RAS proteins sit just inside the cell membrane and relay the signal to grow and divide. A mutation can lock the switch in the "on" position, so the cell keeps dividing when it should stop. Mutated RAS is found in more than 90 percent of pancreatic cancers and in roughly 19 percent of all cancers, which makes it the most commonly mutated cancer gene family known [5].
An obvious drug target, then. The problem was the shape of the protein. Most drugs work by slotting into a pocket on their target, the way a key fits a lock. RAS has an almost featureless surface with no deep pocket for a small molecule to grip. A 2021 review in Signal Transduction and Targeted Therapy summarised the outcome of four decades of effort in one word: undruggable [6].
The first genuine foothold came in 2013. A team at the University of California, San Francisco, led by Jonathan Ostrem, found a small pocket that only appears in one mutant form of the protein, called G12C, and designed compounds that bind there [7]. That discovery eventually produced the first approved RAS drugs. The limitation was built in: the pocket exists only in the G12C variant, so a tumour carrying a different RAS mutation is untouched. In pancreatic cancer, G12C accounts for only about one tumour in a hundred.
How does daraxonrasib work differently from earlier RAS drugs?
Rather than searching for a pocket on each mutant, daraxonrasib borrows a protein that is already inside the cell. It first binds to cyclophilin A, forming a two-part complex. That complex then wraps around the active form of RAS and blocks its ability to pass on the growth signal. The New England Journal of Medicine paper describing the trial calls it a "RAS(ON) multiselective, tri-complex inhibitor" [8].
Because the drug does not depend on the fine detail of one mutation, it can block several common RAS variants at once, including the G12D and G12V forms that dominate pancreatic cancer. That breadth is what made a pancreatic cancer trial possible in the first place.
What did the RASolute 302 trial show?
RASolute 302 enrolled 500 patients across six countries. All had metastatic pancreatic ductal adenocarcinoma that had progressed after one prior line of treatment, a setting where median survival is usually measured in months. Patients were randomly assigned to daraxonrasib tablets or to chemotherapy chosen by their oncologist [8].
The results, published in the New England Journal of Medicine in May 2026, were unusual for this disease:
- Median overall survival across all patients: 13.2 months with daraxonrasib versus 6.7 months with chemotherapy.
- In the RAS G12-mutated group, which made up 92 percent of participants: 13.2 versus 6.6 months, a hazard ratio for death of 0.40.
- Alive at twelve months: 53.3 percent on daraxonrasib versus 18.7 percent on chemotherapy.
- Treatment stopped because of side effects: 1.2 percent on daraxonrasib versus 11.2 percent on chemotherapy [8].
The survival curves separate early and stay apart. The gap between the two median survival figures is about six and a half months, in a cancer where new drugs have historically added weeks.
When the data were presented at the American Society of Clinical Oncology meeting, the audience of oncologists and researchers gave a standing ovation, a reaction rare enough at a scientific conference that footage of it circulated widely [9].
Two caveats belong here. The trial was open-label, so patients and doctors knew which treatment was being given. And the benefit applies to previously treated metastatic disease; trials in earlier lines of treatment are still running.
How fast was the approval, and who could get the drug before it?
The FDA accepted Revolution Medicines' application on July 22, 2026, and approved the drug on August 26. Thirty-five days from acceptance to approval. Daraxonrasib had been selected for the FDA Commissioner's National Priority Voucher pilot, a program designed to compress review times for medicines judged to address a national health priority [1] [10].
Patients did not have to wait even that long. In May 2026, before the application was filed, the FDA cleared an expanded-access program. Expanded access is the legal route through which a seriously ill patient can receive an unapproved drug outside a trial. Within three weeks the company was shipping tablets to treating oncologists, and by early August it had supplied the drug on behalf of more than 2,000 patients through academic cancer centres and community practices across nearly all fifty US states and Puerto Rico [10].
The European Medicines Agency has begun a phased review under its Cancer Medicines Pathfinder project, assessing sections of the application as they are completed rather than waiting for the full dossier [10].
Which other cancers carry RAS mutations?
Pancreatic cancer is the extreme case, but it is not the only one. A 2025 review in Cancer Biology & Medicine put KRAS alterations at roughly 33 percent of lung carcinomas and 50 percent of colorectal cancers, alongside more than 90 percent of pancreatic malignancies [11]. Across every cancer type, the RAS family is mutated in about one patient in five [5].
A phase 3 trial of daraxonrasib in previously treated RAS-mutant non-small cell lung cancer dosed its first patient in May 2025, with enrolment expected to complete in 2026 and a first readout anticipated in 2027 [12]. The FDA has granted the drug Breakthrough Therapy Designation for a subset of these lung cancer patients [10]. Combinations with chemotherapy and immunotherapy are also under study.
Where did cancer immunotherapy begin?
The second half of this story belongs to a different approach: teaching the immune system to attack a tumour. Its origin is usually traced to a New York surgeon in 1890.
That summer, a 17-year-old named Bessie Dashiell trapped her hand between two railway carriage seats. The pain never settled. A biopsy by William Coley, then 28 and newly qualified, found a bone sarcoma. Her forearm was amputated in November 1890. The cancer had already spread, and she died in January 1891 [13].
Coley went back through the hospital's records looking for anything that might have saved her. He found the case of Fred Stein, a German immigrant whose neck sarcoma had regrown after every operation until he caught a severe streptococcal skin infection. As the infection ran its course, the tumour shrank. Coley tracked Stein down in the tenements of the Lower East Side seven years later and found him alive with no sign of cancer [14].
In October 1891, Coley deliberately infected a patient named Zola, whose tonsil tumour was too advanced for surgery. The tumour regressed completely. Zola lived for years before the cancer eventually returned [15] [13]. Coley went on to treat around a thousand patients with a preparation of heat-killed bacteria that became known as Coley's Toxins. Some tumours regressed; many did not, and the unpredictable results drew heavy criticism [16].
For most of the twentieth century, the verdict on Coley was that he had been wrong. The first hard evidence otherwise came in 1976, when a urologist instilled the BCG tuberculosis vaccine directly into the bladders of nine patients with bladder cancer. Fifty years on, intravesical BCG is still the standard adjuvant treatment for high-risk non-muscle-invasive bladder cancer, and a 2025 review in Clinical Microbiology Reviews calls it the most successful cancer vaccine to date [17].
Why did earlier cancer vaccines fail?
BCG was the exception. In 2004, a review by Steven Rosenberg and colleagues at the US National Cancer Institute pooled the results of 1,306 cancer vaccine treatments and found an objective response rate, meaning measurable tumour shrinkage, of just 3.3 percent [18].
The core difficulty is that a cancer cell is a modified version of the patient's own cell. Most of the vaccines in that review pointed the immune system at proteins the tumour shares with healthy tissue. The immune system is trained from birth to ignore those proteins, a safeguard called central tolerance, so the vaccines had little to work with.
A mutation changes that. When a tumour's DNA mutates, it can produce protein fragments that no healthy cell makes. Immunologists call these neoantigens. A 2015 review in Science set out why they matter: because they are new to the body, they escape central tolerance and look to the immune system more like a fragment of a virus than a piece of self [19].
The catch is that every tumour carries its own set of mutations. One patient's neoantigens are not another's. A vaccine built around neoantigens therefore cannot be manufactured in bulk. It has to be designed for one person.
How is a personalized cancer vaccine made?
The process has two computational steps before anything is manufactured.
The first is subtraction. The tumour's DNA is sequenced, the patient's healthy blood cells are sequenced, and every mutation present in the tumour but absent from the blood goes on a candidate list.
The second is prediction. Not every mutant fragment can be displayed on the cell surface for immune cells to see, and a fragment that is never displayed is useless as a target. Machine-learning models predict which candidates will be presented. The 2017 version of the most widely used model was trained on more than 850,000 measured examples of fragment binding; the 2020 version drew on more than 13 million data points covering 250 variants of the presenting molecules [20] [21].
The chosen targets are then encoded into an mRNA vaccine, manufactured for that single patient, and given alongside an immune checkpoint inhibitor.
What did the INTerpath-001 melanoma trial find?
On August 19, 2026, Merck and Moderna announced that INTerpath-001, a phase 3 trial of 1,137 patients, had met both of its endpoints. Participants had stage IIB to IV melanoma that had been completely removed by surgery. Each received either pembrolizumab alone or pembrolizumab plus a personalized mRNA vaccine, intismeran autogene, built from their own tumour's mutations. The vaccine group had longer recurrence-free survival and longer distant-metastasis-free survival [22].
The trial's principal investigator, Georgina Long of Melanoma Institute Australia and joint 2024 Australian of the Year, described it as the first phase 3 study to show that a therapy designed on "the unique mutational 'fingerprint' of a patient's own tumor" can reduce the risk of recurrence or death [22]. Moderna's share price rose 177 percent that day, the largest single-day gain for an S&P 500 company this century.
Melanoma was a deliberate first choice. It is among the most heavily mutated cancers, which gives the vaccine designers the most raw material. Whether the approach transfers to cancers with fewer mutations is the open question, and it is being tested. Merck and Moderna list nine phase 2 and 3 trials of the therapy: two in melanoma, four in lung cancer (three at phase 3), one in kidney cancer and two in bladder cancer. One of the bladder trials pairs the vaccine with BCG, the same tuberculosis vaccine first instilled in 1976 [23].
The wider field is larger than one partnership. A May 2026 analysis in Nature Reviews Drug Discovery mapped the full clinical pipeline of therapeutic cancer vaccines by cancer type and platform, with dozens of candidates in melanoma, lung, glioblastoma, breast and pancreatic cancer [24].
Why did cancer research have a reproducibility problem?
Scepticism about cancer breakthroughs is well earned, and the reason is documented. In 2012, Glenn Begley and Lee Ellis published a commentary in Nature reporting that oncology had the highest clinical trial failure rate of any therapeutic area. They had attempted to confirm the findings of 53 landmark preclinical cancer papers, the kind of studies drug programs are built on. Only six held up [2].
A larger independent effort, the Reproducibility Project: Cancer Biology, published its results in eLife in 2021. It repeated experiments from 23 papers. For the findings that were originally reported as positive, the median effect in the repeat experiments was 85 percent smaller than in the original [3].
Those two papers explain why a cancer cell dying in a dish has so rarely turned into a patient living longer. Naming the problem was the first step in correcting it. Both daraxonrasib and the personalized vaccine were carried through phase 3 trials of 500 and 1,137 patients before anyone called them breakthroughs.
What do these pieces add up to?
Taken separately, each development is incremental. Taken together, they describe a faster pipeline from laboratory to patient:
- The reproducibility problem in preclinical cancer research has been named and is being corrected [2] [3].
- A protein family described as undruggable for four decades now has an approved inhibitor with a survival benefit in phase 3 [6] [8].
- The regulatory process moved from application to approval in 35 days, and expanded access put the drug in patients' hands months earlier [1] [10].
- Personalized cancer vaccines, an idea that traces back to 1890, have their first phase 3 win, with nine further trials underway [22] [23].
- Protein-structure prediction tools such as AlphaFold now model how proteins fold and how drug molecules fit them, which shortens the search for the next target.
None of this is a cure, and the pancreatic cancer figures describe a median of just over a year in a previously treated metastatic population. What has changed is the rate at which genuine results are reaching real patients, rather than stalling in the laboratory. The same tools that produced a RAS inhibitor and a one-patient vaccine are now pointed at lung, colorectal, kidney and bladder cancer, and the readouts are scheduled, not hoped for.
References
2. https://www.nature.com/articles/483531a
3. https://elifesciences.org/articles/71601
4. https://www.nature.com/articles/300149a0
5. https://pmc.ncbi.nlm.nih.gov/articles/PMC7367715/
6. https://www.nature.com/articles/s41392-021-00780-4
7. https://www.nature.com/articles/nature12796
8. https://www.nejm.org/doi/full/10.1056/NEJMoa2605555
9. https://x.com/DrSamuelBHume/status/2061225858384248845
11. https://www.cancerbiomed.org/content/22/7/762
13. https://www.cancerresearch.org/blog/the-legacy-of-bessie-dashiell
14. https://pmc.ncbi.nlm.nih.gov/articles/PMC1888599/
15. https://www.thepharmacologist.org/william-coley
16. https://pmc.ncbi.nlm.nih.gov/articles/PMC7232517/
17. https://journals.asm.org/doi/10.1128/cmr.00194-23
18. https://pmc.ncbi.nlm.nih.gov/articles/PMC1435696/
19. https://www.science.org/doi/10.1126/science.aaa4971
20. https://academic.oup.com/jimmunol/article/199/9/3360/7977122
