Bacterium Inside Liver Tumors May Make Immunotherapy Work Better

If you or someone you love is navigating liver cancer, you have probably already run into the frustrating reality that checkpoint immunotherapy, drugs that unleash the immune system on tumors, works for only a minority of patients. A new mechanistic study offers a surprising reason why some tumors respond and others do not: a bacterium living inside the tumor itself. Researchers at Nanjing Medical University found that Bacillus cereus, a bacterial species that can colonize liver tumor tissue, secretes an enzyme that reorganizes the immune cells around it into a configuration that predicts, and appears to help drive, a stronger response to immunotherapy.

Key takeaways

• Bacillus cereus bacteria were found living inside human liver tumors, concentrated near immune cell clusters called stem-like lymphoid niches (SLNs).
• A bacterial enzyme called PI-PLC triggers a chain reaction that reprograms nearby immune cells, and pairing it with anti-PD-L1 immunotherapy shrank tumors more than either treatment alone in mice.
• This is a preclinical, mechanistic study, done in mouse models and human tumor tissue samples, not a tested human treatment. Nobody should attempt to use bacteria or this enzyme as therapy outside a clinical trial.

Bacterium Inside Liver Tumors Fuels a New Immune Niche

Researchers discovered that a subset of immune cell clusters inside liver tumors, which they call stem-like lymphoid niches (SLNs), are the actual predictor of whether a patient responds to checkpoint immunotherapy, and that a bacterium living inside the tumor helps build them. The study, published in the journal Cell's Cancer Cell, focused on hepatocellular carcinoma (HCC), the most common form of liver cancer. Immunologists already knew that clusters of immune cells inside tumors called tertiary lymphoid structures (TLSs, essentially miniature lymph nodes that form inside the tumor) are generally a good sign for immunotherapy. But the Nanjing team found that not all TLSs are equal: SLNs are a specific, more potent subtype built from CD70-positive type 2 innate lymphoid cells (ILC2s, an immune cell type that helps organize local immune responses) paired with a stem-like population of T cells called Tpex cells (progenitor exhausted T cells, a reserve of cancer-fighting T cells that can keep renewing themselves instead of burning out). Using 16S rRNA sequencing (a technique that identifies bacterial species by their genetic barcode) on tumor tissue from 46 HCC patients, the team found that tumors rich in SLNs also had a distinct microbial signature, and singled out Bacillus cereus as enriched in SLN-rich tumors. Fluorescence imaging on a separate set of 32 patient tumors confirmed the bacteria were physically colocalized with SLNs, and the researchers cultured live, viable B. cereus directly out of tumor tissue.

Why Liver Cancer Immunotherapy Often Fails Patients

Liver cancer is one of the world's most lethal cancers, and hepatocellular carcinoma accounts for the large majority of cases. Checkpoint immunotherapy drugs that block PD-1 or PD-L1 (proteins that cancer cells use to hide from immune attack) have become a standard option, but response rates in HCC remain limited, with a large share of patients seeing no benefit. This study helps explain part of that gap. Across 50 HCC patients evaluated at two clinical centers, tumors rich in TLSs generally responded better to anti-PD-1 therapy, but a notable subset, 13 of those TLS-rich cases, still failed to benefit. Looking across multiple cancer types, the researchers found that more than half of patients with TLS-rich tumors did not respond to PD-1/PD-L1 blockade. Even the "mature" version of TLS that doctors have generally considered the most favorable sign, ones containing germinal centers, was misleading on its own: 57% of patients with these mature TLS were still immunotherapy-resistant, compared with 17% of patients whose TLS lacked germinal centers. In other words, doctors have been reading a fuzzy signal. SLNs appear to be the sharper one, and the bacterium living nearby seems to be helping build it.

How Researchers Traced the Bacteria to Human Tumor Tissue

The team combined human tissue analysis with tightly controlled mouse experiments to nail down cause and effect, not just correlation. Beyond the imaging and sequencing in patient samples, they used orthotopic mouse liver tumor models (Hepa1-6 and H22, tumors grown directly in the mouse liver rather than under the skin, to better mimic the real organ environment) and directly administered B. cereus to see what happened. They then used genetically engineered mice lacking ILC2s, mice lacking the alarm-signal protein IL-33 (an "alarmin," a molecule tissue releases when stressed or damaged that alerts nearby immune cells), and a lab-engineered strain of B. cereus missing the gene for its PI-PLC enzyme (phosphoinositide-specific phospholipase C, an enzyme that cuts up a component of cell membranes) to see which piece of the chain was doing the work. Removing any one of these components, the immune cells, the alarm signal, or the bacterial enzyme, blunted the effect. The team also ran experiments in germ-free mice (raised with no microbes at all) and found that PI-PLC alone, without any live bacteria present, was enough to trigger the same immune reprogramming. That is what let them call PI-PLC the actual effector molecule, rather than just a bystander.

What This Means If You or a Loved One Faces Liver Cancer

For now, this changes nothing about how liver cancer is actually treated. What it does is sharpen the picture of why immunotherapy works for some patients and not others, and it points toward two future possibilities worth watching: SLN status as a way to predict who is likely to benefit from checkpoint immunotherapy before starting it, and PI-PLC, or a drug designed to mimic its effect, as a potential add-on to boost immunotherapy in patients whose tumors lack a strong SLN signature. In mice, combining PI-PLC treatment with anti-PD-L1 antibody increased tumor-killing CD8+ T cell activity and shrank tumors more than the checkpoint drug alone. None of that has been tested in people. If you are weighing immunotherapy options for liver cancer, this study is a reason to ask your oncologist about biomarker testing and clinical trial availability, not a reason to pursue any particular supplement, probiotic, or bacterial exposure. There is no evidence that ingesting or otherwise trying to acquire B. cereus would replicate anything seen in this study, and B. cereus is also a known cause of foodborne illness in other contexts, so self-experimentation is not just unproven but potentially unsafe.

The Open Questions This Preclinical Study Leaves Unanswered

The biggest open question is how the bacteria get into the tumor in the first place and whether that route can be predicted or manipulated safely. The authors note that the liver's unique blood supply, it receives blood directly from the gut via the portal vein, may make it especially exposed to gut-derived microbes, but this study did not map that migration route in detail. It is also not yet clear whether boosting B. cereus or PI-PLC exposure in a real patient would reliably help rather than trigger unwanted inflammation, since the bacterium can behave as an opportunistic pathogen outside this specific tumor context. The patient cohorts, while analyzed with sophisticated spatial and single-cell tools, came from two centers in China, and the mouse experiments used group sizes in the range of 6 to 9 animals per group, typical for this kind of mechanistic work but not the scale needed to establish real-world clinical benefit. Independent replication in larger, more diverse patient populations has not yet happened.

What Comes Next Before This Reaches the Clinic

The logical next steps are validating SLN status as a clinical biomarker in larger, multi-center patient cohorts, and testing whether a PI-PLC-based or IL-33-pathway-based drug can be developed and safely dosed in humans, likely starting with early-phase safety trials rather than direct patient benefit studies. The authors frame SLNs and the underlying bacterial-immune axis as a therapeutic target worth pursuing, but that pursuit is still in its early, preclinical stages, and any human trial would likely be years away.

Common Questions

Is this a new cancer treatment I can get now? No. This is laboratory and mouse research plus analysis of existing human tumor samples. No drug or bacterial therapy based on this finding has been tested in a clinical trial yet.

Should I try to get more Bacillus cereus in my diet or gut? No. The study did not test oral or dietary exposure, and B. cereus is a known cause of food poisoning outside this specific tumor context. There is no evidence this would help and it could be harmful.

What is a "stem-like lymphoid niche" in plain terms? It is a specific cluster of immune cells inside a tumor, built from two cell types working together, that the study found predicts and may help drive a stronger response to checkpoint immunotherapy in liver cancer.

Does this apply to cancers other than liver cancer? The mechanism was studied specifically in hepatocellular carcinoma. The researchers note that TLS-related immunotherapy resistance shows up across multiple cancer types, but the bacterial mechanism described here has only been demonstrated in liver tumors.

References

1. Chupeng Hu, Wenhua You, Yedi Huang, et al. (2026). Intratumoral Bacillus cereus orchestrates stem-like lymphoid niches to potentiate immunotherapy response in hepatocellular carcinoma. Cancer Cell. https://doi.org/10.1016/j.ccell.2026.07.016
2. Qianru Li, Chao Ding, Maomao Cao, et al. (2024). Global epidemiology of liver cancer 2022: An emphasis on geographic disparities. Chinese Medical Journal. https://doi.org/10.1097/cm9.0000000000003264
3. Qian Li, Jingjing Han, Yonglin Yang, Yu Chen (2022). PD-1/PD-L1 checkpoint inhibitors in advanced hepatocellular carcinoma immunotherapy. Frontiers in Immunology. https://doi.org/10.3389/fimmu.2022.1070961