A New CAR T-Cell Therapy Just Shrank Solid Tumors in Mice. Pancreatic Cancer Researchers Are Cautiously Optimistic

By all accounts, one of the most depressing oncology diagnoses is pancreatic cancer. The most prevalent type, pancreatic ductal adenocarcinoma, has a five-year survival rate that is persistently below 13%. The median survival time for patients with metastatic disease is still less than a year. Not much has changed despite decades of research, numerous clinical trials, and chemotherapy regimens. Immunotherapy, which has shown impressive results in other cancers, is especially ineffective against the disease. Therefore, even though the word “mice” is doing a lot of silent work in that sentence, a study from Memorial Sloan Kettering that reports that a novel form of CAR T-cell therapy shrank pancreatic tumors and cleared metastases in laboratory models garners attention.

The study, which was published in Cell in late March 2026, is an example of the kind of preclinical outcome that scientists have been attempting to achieve for years with varying degrees of success. In cases of blood cancers, CAR T-cell therapy—a strategy that entails modifying a patient’s own immune cells to identify and combat cancer—has been truly revolutionary. In certain trials, complete remission rates following CAR T treatment for lymphoma and leukemia have reached 80% to 90%. The technology is functional. The issue has always been that solid tumors are a completely different biological environment, and when CAR T cells come into contact with a tumor embedded in dense, fibrous tissue surrounded by a network of immunosuppressive cells that actively block the immune response, the elements that make the cells so effective in the bloodstream tend to disintegrate.

The MSK team tackled this issue from a different perspective than most previous attempts, under the direction of postdoctoral researcher Zeda Zhang and co-senior authors Scott Lowe and Michel Sadelain. They created CAR T cells that target uPAR, a surface protein that is present on both tumor cells and the supporting architecture surrounding the tumor, as opposed to the traditional strategy of designing CAR T cells to target the cancer cells themselves, which keeps running into problems with target variability and immune suppression. Fibroblasts, immune-suppressive myeloid cells, and scar-tissue builders all contribute to the cancer’s successful fortification. According to Lowe, uPAR identifies not only the cancer itself but also the larger ecosystem that fosters cancer growth.

It may not seem important, but that distinction is crucial. The architectural nature of solid tumors has contributed significantly to their resistance to immunotherapy. The tumor recruits in addition to growing. It creates a microenvironment that actively repels immune cells by coopting the surrounding tissue’s natural wound-healing mechanisms. The majority of earlier CAR T strategies have attempted to break through that barrier. In addition to focusing on what’s behind the wall, this strategy attempts to destroy the wall itself. uPAR-targeting CAR T cells not only reduced the size of the tumor in a mouse model of ovarian cancer, but they also eliminated metastases and created long-lasting remissions. The engineered T cells remained active enough to fend off subsequent attempts by researchers to introduce fresh cancer cells into the same mice.

Across a variety of cancers examined in the study, the uPAR results were consistent. Twelve of the fourteen human cancer types that were examined had elevated uPAR expression, with pancreatic, ovarian, lung, colon, and some brain cancers having especially high levels. The protein seems to be particularly active in tumors with activating mutations in KRAS, a gene pathway that is, ironically, one of the most frequent causes of pancreatic cancer, and p53 mutations, the tumor-suppressor gene sometimes referred to as the guardian of the genome. To put it another way, the cancers that are most visible to this specific approach may also be the ones that are most difficult to treat.
There is another noteworthy discovery. When uPAR-targeted CAR T cells were combined with chemotherapy drugs that cause cellular senescence, such as cisplatin, the researchers found that the therapy was more successful because the chemotherapy forced more cancer cells into a state that increased uPAR expression on their surface, thereby painting more targets for the engineered cells to find and attack. It’s unclear if this combination will be tolerable and have manageable side effects in people. However, it points to a possible way to improve the treatment beyond what CAR T cells can accomplish on their own.
It’s important to be truthful about the current situation. This research is preclinical. People are not mice. Results that appeared promising in mouse models but failed to translate are numerous in the history of cancer research. Targeting antigens like mesothelin, PSCA, and MUC1, CAR T therapy for pancreatic cancer has been in early clinical testing for years with, at best, modest results, and the tumor microenvironment is still an unresolved issue. There is real cause for cautious optimism as this research advances, but there is also the familiar burden of knowing how many promising preclinical discoveries have stalled somewhere between the lab and the clinic.

Even so. By 2030, pancreatic cancer is expected to rank as the second most common cause of cancer-related deaths in the US. There are still incredibly few options available. It is not insignificant that a research team at one of the top cancer centers in the world has discovered a target that seems to target the tumor’s support structure rather than just the tumor itself, and that the strategy was successful in preclinical models for a variety of cancer types. It’s not the kind of outcome that warrants a press release referring to it as a cure, but rather one that warrants cautiously approaching human trials. That is precisely what the researchers themselves have been cautious to state. Testing for efficacy and safety in humans is the next stage. We haven’t started that work yet.