Pancreatic cancer:
Blocking its growth.
Teaching the immune system to find it.

Pancreatic cancer is beginning to yield to a more precise understanding of its biology. Scientists can now interfere with a growth signal that long resisted drug development. They are also learning how to train immune cells to recognize a patient's particular tumor. These advances address different problems, and the evidence behind them is at different stages. Understanding both explains why the progress matters—and why there is still so much work to do.

This first Therapeutic Perspective examines daraxonrasib, an oral medicine developed by Revolution Medicines, and autogene cevumeran, an individualized mRNA vaccine being developed by BioNTech and Genentech. Our focus is pancreatic ductal adenocarcinoma, or PDAC, the most common form of pancreatic cancer. Other pancreatic tumors, including neuroendocrine tumors, have different biology and should not be folded into the same analysis. [1]

Why this cancer has been so difficult

The pancreas sits deep in the abdomen and helps with digestion and blood-sugar regulation. A tumor can grow before it creates a distinctive warning. By diagnosis, cancer may already have spread beyond the organ. Metastatic means it has established disease at distant sites; resectable means surgeons can remove the tumor. These are profoundly different starting points. Even after apparently successful surgery, microscopic cancer cells can remain. Treatment given afterward, called adjuvant therapy, tries to eliminate those cells before they regrow. [1]

A pancreatic tumor is also more than a collection of malignant cells. It contains connective tissue, immune cells, blood vessels and other supporting structures. This surrounding environment—the stroma—can influence drug delivery and immune activity. Dense tissue is part of the problem, but the picture is more complex than a wall that simply needs knocking down: interactions among different cells and signals matter. Research on abnormal collagen, a structural protein, illustrates how changes in this environment can affect both tumor growth and immune exclusion. Much of that mechanistic work is in experimental models, not proof that a particular treatment improves human survival. [2]

Finally, immune cells need more than permission to attack. They need a recognizable target and effective instructions. Pancreatic cancers generally provide fewer mutation-derived targets than highly mutated cancers such as some lung cancers and melanomas. Research on dendritic cells—the immune system's antigen-presenting cells—shows how failures in the recognition process can contribute to weak immune responses. Removing an immune brake alone may accomplish little if there is no effective response to release. [3]

The growth switch scientists struggled to reach

Genes are instructions; proteins carry out much of the work. KRAS is a gene that encodes one member of the RAS protein family. RAS helps relay messages telling a cell when to grow and divide. A mutation changes the instructions and can produce a protein that sends growth signals inappropriately.

RAS normally cycles between an active state, carrying a molecule called GTP, and an inactive state, carrying GDP. Cancer-driving mutations can disrupt the mechanisms that turn signaling off. The familiar image of an accelerator stuck down is useful, although individual mutations do not behave identically. The practical problem is excessive signaling through pathways that help a tumor survive and grow. [4]

RAS-family mutations occur in more than 90% of pancreatic cancers, making this a central biological problem rather than a rare exception. For years, scientists understood that RAS was important but struggled to make a useful drug bind it. A biological target is not automatically a druggable target. Its shape, chemistry and normal function all affect whether a medicine can grip it and disrupt the desired interaction. RAS offered no obvious easy attachment point. Structural research and increasingly sophisticated chemistry gradually changed what was possible. [5]

Daraxonrasib: obstructing the signal

Daraxonrasib takes an unusual route. It binds a protein already present in cells, cyclophilin A. The resulting pair binds active RAS. Together, the drug and the two proteins form a three-part complex that blocks RAS interactions with downstream signaling partners. Think of assembling a clamp around a difficult connection, rather than finding a conventional socket for a drug to plug into. This is an analogy for the interaction, not the molecule's literal shape. [6]

The term RAS(ON) refers to the active, GTP-bound state. Multiselective means the drug can inhibit multiple RAS forms rather than just one particular mutation. It acts against both mutant and non-mutant, or wild-type, RAS. That breadth may be useful, but it is not the same as acting only on cancer cells. Normal tissues also depend on signaling, so tolerability remains important. [6]

The clinical importance is now supported by a randomized comparison. RASolute 302 enrolled 500 patients with metastatic pancreatic cancer after progression following one prior systemic treatment. Patients were assigned to daraxonrasib or a physician-selected standard chemotherapy regimen. In the overall population, median overall survival was 13.2 months versus 6.7 months. Median progression-free survival was 7.2 versus 3.6 months. These are results from the second-line setting, not estimates of survival from initial diagnosis. [7]

Overall survival measures time until death from any cause. Progression-free survival measures time until the cancer worsens or the patient dies. A median is the midpoint of the estimated survival distribution; it is not a deadline for an individual patient. The difference between these medians is not an additional amount of life that every patient is guaranteed to receive.

The FDA approved daraxonrasib on August 26, 2026, for adults with metastatic pancreatic adenocarcinoma who have received at least one prior systemic therapy or are not candidates for multiagent systemic therapy. The pivotal randomized evidence described above came from previously treated patients; the label and the trial population should not be treated as interchangeable descriptions. [16]

Meaningful benefit does not eliminate treatment burden. The label identifies skin toxicity, mouth inflammation, diarrhea and other potentially serious risks, including gastrointestinal perforation and lung inflammation. A daily tablet can simplify administration while still requiring close clinical monitoring. [6]

Why blocking a driver does not necessarily finish the job

Tumors evolve under treatment. Cells that remain dependent on the blocked signal may die or stop growing, while cells with alternative survival routes can become more prominent. Different cells within the same cancer need not behave identically.

Studies of KRAS inhibition have identified several possible routes to resistance. Experimental work combining chemotherapy with a different KRAS inhibitor, MRTX1133, found complementary effects in mice. That supports a rationale for combination research, but it does not establish that the same combination, dose or benefit translates to daraxonrasib in people. The human questions are whether adding treatment delays resistance, improves survival and remains tolerable. [8]

Daraxonrasib is also being studied in earlier treatment settings. Its second-line results create a reason to test earlier use; they cannot substitute for those trials. A drug's best place in care depends on the population, comparison treatment, toxicity and durability—not simply on whether its mechanism is attractive. [9]

The vaccine: teaching the immune system what to look for

A tumor originates from a person's own cells. The immune system therefore faces a recognition problem: how can it distinguish malignant cells from healthy tissue?

Some tumor mutations create altered protein fragments called neoantigens. “Neo” means new; an antigen is something the immune system can recognize. These mutation-derived fragments can supply a distinguishing feature. The purpose of a therapeutic vaccine is to direct an immune response toward such features. It is not a vaccine given routinely to healthy people to prevent pancreatic cancer. [10]

Autogene cevumeran is individualized. Scientists analyze tumor and normal material, identify expressed mutations and select targets for that patient. The vaccine supplies messenger RNA, or mRNA: temporary instructions cells can read to make the selected protein material. Antigen-presenting cells then display fragments to T cells, helping start a targeted immune response. The vaccine does not contain a living pancreatic tumor. [11]

This process has several distinct hurdles. The selected mutation must produce an appropriate target. Cells must display it in a form the person's immune system recognizes. Relevant T cells must multiply, reach cancer cells and remain functional. “We can manufacture the vaccine” and “the vaccine prevents recurrence” are very different achievements.

The pancreatic phase 1 program combined surgery, the immune checkpoint drug atezolizumab, autogene cevumeran and modified FOLFIRINOX chemotherapy. The vaccine encoded up to 20 selected neoantigens. A 2025 peer-reviewed follow-up reported persistent, functional vaccine-induced CD8 T cells in responding patients approximately three years after vaccination. CD8 T cells can directly kill cells displaying the relevant target. Persistence matters because a brief immune response may disappear before residual cancer does. [12]

Atezolizumab blocks PD-L1, part of a signaling system that can restrain immune activity. The proposed logic is complementary: the vaccine helps teach target recognition, while checkpoint inhibition can reduce a brake on the response. Chemotherapy addresses cancer through another mechanism. The clinical value of this combined approach must be established as a regimen; the design does not let us attribute every outcome to the vaccine alone. [13]

Modified FOLFIRINOX is a combination of folinic acid, fluorouracil, irinotecan and oxaliplatin, with adjustments to the original regimen. It is an active treatment, not a placebo. The vaccine strategy therefore has to show what it adds to established care. Its overall safety and treatment burden also need to be evaluated across the complete combination; a small early vaccine study cannot rule out uncommon harms. [13]

What the early vaccine results tell us—and what they cannot

Sixteen patients received the vaccine in the phase 1 study. Eight developed the measured vaccine-induced immune response. In MSK's April 2026 report of conference follow-up, seven of those eight were alive four to six years after surgery, compared with two of the eight without that response. These are small patient counts, reported by the institution conducting the research, rather than results from a completed randomized efficacy trial. [14]

The association deserves attention. But patients were not randomly assigned to “immune responder” and “non-responder” groups. Those categories emerged after treatment and may reflect differences in biology or health as well as treatment effects. Everyone also underwent surgery and received other components of the regimen. Consequently, these observations cannot establish how many deaths the vaccine prevented.

It would be especially misleading to compare these patients with an all-stage pancreatic cancer survival statistic, or with the metastatic daraxonrasib population. People eligible for surgery begin from a different clinical position. Similarly, being alive does not necessarily mean being free of recurrent cancer. Survival, recurrence and immune response are separate measurements.

The randomized phase 2 study, NCT05968326, compares autogene cevumeran plus atezolizumab and modified FOLFIRINOX with modified FOLFIRINOX alone after resection. This is the kind of comparison needed to learn whether adding the immune-based regimen improves clinical outcomes. It tests the added package, not the vaccine in isolation. [13]

Where the approaches compete—and where they do not

These therapies compete for research attention, development resources and a place in future treatment strategies. In the settings reviewed here, they do not answer the same immediate clinical question. One has been tested against an established treatment for visible metastatic disease. The other aims to improve the odds that cancer does not return after removal.

Sources: regulatory and trial materials [6, 7, 12, 13]. This is a comparison of mechanisms and development evidence, not a head-to-head treatment ranking.

For commercial and clinical diligence, we would ask different questions of each program. For daraxonrasib: how durable is benefit, how often does toxicity interrupt treatment, and what do earlier-line or combination trials add? For the vaccine: how many patients can complete the manufacturing and treatment sequence, how consistently is an immune response induced, and does that response translate into outcomes that matter to patients? Manufacturing success and immune activity are necessary operational or biological milestones; neither replaces evidence of clinical benefit.

Why progress is happening now

Our interpretation is that several capabilities have begun to meet. Structural biology and chemistry have made a once-elusive signaling target accessible. Tumor sequencing can identify individual mutations. Computational target selection and mRNA manufacturing can turn that information into an intervention. Immune monitoring can then test whether the intended cells actually appear and persist. These capabilities make hypotheses testable with a precision that a broad statement such as “stimulate the immune system” does not provide. [5] [11] [12]

Timing also changes the scientific problem. Treating extensive, rapidly progressing disease is different from targeting possible microscopic remnants after surgery. A vaccine strategy may have a more favorable task when the amount of cancer is small, but that is a rationale to test—not proof that the strategy succeeds. [15]

Progress becomes more credible when we can connect a biological mechanism to a measured effect—and then connect that effect to a better outcome for patients.

Our assessment and the evidence we will revisit

Daraxonrasib establishes the stronger clinical claim today. A randomized survival benefit provides much firmer evidence than an attractive mechanism or a small uncontrolled series. Its advance is meaningful even though advanced pancreatic cancer remains a life-threatening disease and resistance remains an important problem.

The vaccine supplies a different kind of progress. Sustained, tumor-directed immune responses in some patients challenge the assumption that pancreatic cancer cannot be made visible to an effective immune attack. The outstanding question is whether that biological achievement produces a reproducible benefit across a larger, appropriately controlled population.

We would strengthen our assessment of RAS inhibition if further randomized studies show durable benefit in additional settings without an unacceptable treatment burden. We would strengthen our assessment of the vaccine if randomized results demonstrate a clinically meaningful reduction in recurrence, followed by credible survival evidence. We would become more cautious if immune activity fails to translate into better outcomes, if manufacturing limits practical delivery, or if toxicity substantially offsets benefit.

Combining direct tumor inhibition with immune targeting is an appealing hypothesis. These separate studies do not establish that daraxonrasib and autogene cevumeran should be combined, or how to sequence them. Any such strategy needs its own evidence.

A short glossary

PDAC

Pancreatic ductal adenocarcinoma, the pancreatic cancer type examined here.

Systemic treatment

Treatment that acts throughout the body, rather than only at a local site.

Second-line

The next treatment after an initial treatment has failed or cannot be continued.

Neoantigen

A new molecular target generated by a tumor mutation that immune cells may recognize.

Recurrence

Cancer returning after a period in which it was not detected.

Phase 1 / 2 / 3

Broad stages of clinical development: early safety and biological testing; further efficacy testing; and larger confirmatory comparisons. The actual design matters more than the phase label alone.

Randomized trial

A study assigning treatment by chance to reduce systematic differences between comparison groups.

Therapeutic Perspectives provides general scientific and competitive analysis. It is not medical advice, an individual treatment recommendation or investment advice. Clinical decisions depend on a patient's circumstances and qualified medical care. Views reflect the evidence cutoff and may change with new data.

Return to the Therapeutic Perspectives library →