The paper made the rounds this month as a shield-breaker. In the second week of September, riding an $800,000 grant announcement out of the University of Miami, word went out that scientists had found a way to break pancreatic cancer’s protective shield. What the coverage did not lead with sits second-to-last on the study’s fourteen-name author list: David Liberg, chief scientific officer of Cantargia AB, the Swedish biotech that makes the drug the study is about.
It is a good image, that shield, and pancreatic cancer earns it. The tumor comes wrapped in so much scar tissue and so many immune-suppressing cells that chemotherapy and immunotherapy both tend to bounce off, and anyone who has watched the disease work knows why a shield-breaker would be worth writing up. The study itself had come out in June, nearly three months earlier, in JCI Insight, with Jashodeep Datta, the Sylvester Comprehensive Cancer Center surgeon running the program, as its corresponding author. The disclosures note that Datta himself receives funding from Cantargia and consulting fees from Boston Scientific. It is all printed, and it is exactly the kind of thing that does not survive the trip from the journal to the press release.
So before the shield gets any bigger, read what the record shows. The target is a receptor called IL1RAP, a shared docking point that several inflammatory signals lean on. Block it, the theory goes, and you cut the wiring that lets tumor cells, fibroblasts, and immune cells cooperate to keep the drugs out. To test that, Datta’s team bred pancreatic tumors into genetically engineered mice and treated them with a murine version of Cantargia’s antibody, nadunolimab, also known as CAN04.
The mice did better. Tumor growth was restrained (P=0.019), the scarring around the tumors thinned measurably (P=0.0003), suppressive myeloid cells fell, and exhausted T cells shifted back toward a fighting state. The four-drug regimen, the antibody plus gemcitabine, paclitaxel, and an anti-PD1 checkpoint drug, pushed median survival to 54 days. The control mice reached 43. The gap clears statistical significance handily (P<0.001), and it is eleven days in a mouse.
That mouse experiment is the paper’s original work. Everything it says about humans is reanalysis. The team ran the antibody’s fingerprints back through transcriptomic data from the COMPASS cohort (195 patient samples), a single-cell atlas of nearly 655,000 cells, and tumor biopsies from 49 patients already enrolled in Cantargia’s own human trial. IL1RAP tracked with worse survival, and patients given nadunolimab showed fewer suppressive myeloid cells and more active CD8 T cells afterward. That last, before-and-after immune shift, the one that most looks like the drug doing something in a person, rests on paired biopsies from two patients. In the authors’ own words, these are post hoc analyses of trials that were run for other reasons.
The human trial is CANFOUR (NCT03267316), an early-phase study of nadunolimab with chemotherapy in advanced pancreatic cancer, and it is where the market-moving number lives. Across all patients, the combination produced a median overall survival of 13.2 months, respectable but drawn from an uncontrolled trial measured against historical benchmarks rather than a control arm. The eye-catching figures, a 48 percent response rate and a 35 percent two-year survival, belong to the subgroup whose tumors expressed high levels of IL1RAP. Those subgroup numbers are what Cantargia points to around the drug’s FDA fast-track designation, and the company is now building a companion test to pick IL1RAP-high patients out in advance. Fast track is a promise to review quickly, not a verdict that the drug works, and a survival number pulled from the biomarker-defined slice of an uncontrolled trial is a hypothesis wearing a headline’s clothes.
Which is the honest shape of the thing. A company with one lead asset has a plausible mechanism, encouraging early data, and a biomarker that conveniently sorts its best responders into a group that looks great on a slide. Its chief scientific officer helped write the paper that dignifies the mechanism. Its academic collaborator, funded by the company, is now moving the drug into a first-of-its-kind neoadjuvant trial, giving nadunolimab to people with operable pancreatic tumors before surgery, paid for by an $800,000 V Foundation grant. None of that is misconduct. All of it is how a biotech turns a receptor into a revenue line, and the disclosures sit there in the fine print precisely because someone anticipated the question.
What is missing is the only thing that would settle it: a controlled trial, in people, where nadunolimab is compared head-to-head against chemotherapy alone and the IL1RAP-high subgroup is named in advance rather than found afterward. Cantargia says it is preparing a randomized, potentially pivotal study. Until it reports, the shield metaphor is doing the work the data cannot.
Datta called the move into the clinic a landmark for his cancer program, and he may turn out to be right. Pancreatic cancer has broken more promising mechanisms than anyone can count, and it will not be embarrassed by a press release. In the one comparison that had a control arm, the mice, the fully loaded regimen bought eleven days. The trial that would tell him whether it buys a person anything is, by the company’s own account, still being built.
Sources
- JCI Insight – Dickey, Marsh, Liberg, Datta et al., “IL1RAP-expressing myeloid-stromal networks… in pancreatic cancer” (June 22, 2026)
- PMC full text – funding and conflict-of-interest disclosures
- CancerNetwork – Nadunolimab earns FDA fast-track designation in metastatic PDAC (subgroup response and 2-year survival)
- CancerNetwork – Nadunolimab combination efficacy in advanced/metastatic PDAC (CANFOUR, 13.2-month median OS)
- Cantargia AB – Updated CANFOUR survival data; randomized pivotal trial in preparation
- University of Miami / InventUM – V Foundation grant and planned neoadjuvant trial
- ScienceDaily – “Scientists find a way to break pancreatic cancer’s protective shield” (Sept. 2026)