Gut Bacteria May Sabotage CAR-T Cancer Therapy, Scientists Find a Fix

If you or someone you love is facing pancreatic cancer, you've likely heard about CAR-T therapy, a treatment that reprograms a patient's own immune cells to hunt down tumors, and you may have also heard that it barely works against this particular cancer. A new laboratory study offers one explanation why: bacteria living in the gut of pancreatic cancer patients appear to quietly sabotage the very immune cells doctors are trying to weaponize. The research, posted as a preprint by a team at Harbin Medical University's First Affiliated Hospital, traces the sabotage to a single enzyme, and points toward a possible fix.

Key takeaways

• Gut bacterial metabolites, whether from cancer patients or healthy volunteers, shifted CAR-T cells toward a more suppressive profile in lab dishes, but only metabolites from pancreatic cancer patients significantly weakened the cells' actual tumor-killing power, with the sharpest drop at a 1:1 ratio of immune cells to tumor cells.
• The sabotage traced back to cholesterol metabolism inside the CAR-T cells; bacterial metabolites from cancer patients pushed up cholesterol-building genes, including one called ACAT-1.
• Blocking ACAT-1 reversed the suppression entirely; CAR-T cells kept their killing power even when exposed to cancer-patient gut bacteria, and in some tests killed better than untreated controls.
• This is an early, unpublished preprint built on 9 pancreatic cancer patients versus 20 healthy controls, and every CAR-T experiment happened in a dish, not in a living tumor, so none of this is ready to guide treatment yet.

Gut Bacteria Blunt CAR-T Killing Power

Gut bacteria from pancreatic cancer patients directly weakened the cancer-killing ability of CAR-T cells (chimeric antigen receptor T cells, immune cells genetically engineered to recognize and attack a specific target on tumors) in the new preprint from a team led by Lei Zhao and colleagues. The researchers filtered stool samples from patients and healthy people down to a bacteria-free supernatant, essentially the metabolic byproducts bacteria leave behind, and mixed it with CAR-T cells engineered to target mesothelin (MSLN), a protein commonly overexpressed on pancreatic tumor cells. Supernatant from either group, patients or healthy volunteers, shifted the CAR-T cells toward a more immunosuppressive profile, with a rise in Treg cells (regulatory T cells, a subtype that dampens immune attacks) and a drop in TH1 cells (a subtype that drives inflammation against threats). But when the team tested actual tumor-killing at an effector-to-target ratio of 1:1, meaning equal numbers of CAR-T cells and tumor cells, only the cancer patients' gut bacteria produced a significant drop in killing power; at higher ratios of 5:1 and 10:1, the difference disappeared.

Pancreatic Cancer Needs Better Immunotherapy

Pancreatic cancer needs new immunotherapy options because current treatments fail more patients than almost any other major cancer. It kills more than 400,000 people worldwide each year, and more than half of patients are already unresectable, meaning surgery isn't an option, by the time they're diagnosed. Even CAR-T therapy engineered specifically to target mesothelin, which showed early promise shrinking abdominal metastases in a handful of patients in trials led by Carl June's group, has stalled for years without major progress. The gut microbiome's broader influence on immunotherapy outcomes is well documented in other cancers, and pancreatic cancer's dismal outlook, roughly 3% five-year survival once the disease has spread, according to the American Cancer Society, makes any lever worth pulling.

Small Study, Careful Multi-Step Design

The credibility of this finding rests on a multi-step design that moves from real patients to lab validation. The team started with fecal samples from 9 pancreatic cancer patients and 20 healthy controls, sequencing gut bacteria with 16S rRNA analysis and finding the two groups' bacterial communities were markedly different: healthy volunteers carried far more distinct bacterial genera (3,780) than patients did (664), with Streptococcus and Klebsiella notably more abundant in the cancer group and Faecalibacterium and Bifidobacterium more abundant in healthy volunteers. From there, the researchers ran metabolomics on the same samples, comparing more than 200,000 metabolite signatures, before functionally testing the bacterial supernatants on live CAR-T cells and finally using RNA interference (a technique that silences a specific gene) to knock down ACAT-1 and test whether removing it reversed the damage. ACAT-1 was already known, from prior research on CD8+ T cells, as a rate-limiting enzyme in T-cell cholesterol metabolism.

What This Means If You Are Watching CAR-T Trials

For now, this research does not change what a pancreatic cancer patient or caregiver should do today; it's a mechanistic clue, not a treatment. If you're researching CAR-T options, it may be worth asking your oncology team whether any trials are exploring microbiome status or cholesterol-pathway modulation alongside cellular therapy. It's worth being explicit: nothing here validates probiotics, specific diets, or antibiotics as a way to improve CAR-T outcomes today. The authors speculate about future strategies such as suppressing certain bacteria found elevated in patients or supplementing others more abundant in healthy volunteers, but they present these as ideas for future research, not recommendations, and none of it has been tested in humans.

What This Study Cannot Yet Prove

This study cannot yet prove that fixing gut bacteria, or blocking ACAT-1, will help a real patient's tumor shrink, because every CAR-T experiment here happened in a dish, not in an animal or a person with an actual tumor. The paper is also an unpublished preprint posted to ResearchSquare that has not been peer reviewed, and its human comparison is small: 9 pancreatic cancer patients against 20 healthy controls. The researchers themselves acknowledge they tested only the combined effect of each group's full bacterial supernatant rather than isolating which individual bacterial species or metabolites are doing the damage.

What Comes Next for This Research

The authors say their next step is narrowing down which specific bacterial species and individual metabolites, rather than the whole bacterial soup, are responsible for suppressing CAR-T cells, and testing whether reshaping the gut microbiome through antibiotics or targeted probiotic supplementation changes CAR-T outcomes in a living tumor model. No timeline was given.

Common Questions

What is MSLN-CAR-T therapy? MSLN-CAR-T is CAR-T cell therapy engineered to target mesothelin (MSLN), a protein commonly overexpressed on pancreatic cancer cells; it has shown some ability to shrink abdominal metastases in early trials but has not become a reliable pancreatic cancer treatment.

What is ACAT-1 and why is it a drug target? ACAT-1 is an enzyme that helps cells store cholesterol; blocking it appears to free up cholesterol in the T-cell membrane in a way that strengthens the immune cell's ability to organize and kill tumor cells.

Does this mean gut bacteria cause pancreatic cancer? No. This study did not test whether gut bacteria cause pancreatic cancer; it tested whether bacterial metabolites from existing patients interfere with CAR-T therapy after the cancer is already present.

Should I change my diet or take probiotics based on this study? No. The study did not test any diet or probiotic intervention in humans, and the authors describe microbiome-targeted strategies only as ideas for future research, not as something proven to help.

References

1. Lei Zhao, Zixian Wei, Tao Zhang, et al. (2026). Gut Microbiota inhibit the potency of MSLN-CAR-T cells against pancreatic carcinoma by modulating Cholesterol Metabolism. https://doi.org/10.21203/rs.3.rs-10451699/v1
2. Nature Reviews Cancer: microbiota and cancer therapy. Nature. https://www.nature.com/articles/nrc.2017.13
3. Nature: cholesterol metabolism and CD8+ T cell antitumour response. Nature. https://www.nature.com/articles/nature17412