Gut Microbe Transplants May Reverse Immunotherapy Resistance in Liver Cancer

One of the most frustrating realities of modern cancer medicine is that even the most celebrated immunotherapies stop working in a large fraction of patients. For advanced hepatocellular carcinoma, the most common form of liver cancer, the standard first-line treatment combines the immune checkpoint inhibitor atezolizumab with the blood-vessel-blocking agent bevacizumab. Yet many tumors eventually progress despite this combination, leaving clinicians with few good options. A new phase II pilot study published in Nature Communications suggests an unexpected ally in this fight: the trillions of bacteria living in the human gut. Researchers at the Medical University of Vienna report that a single fecal microbiota transplantation, delivered alongside a renewed course of the same drug combination, was safe and associated with tumor shrinkage in a striking proportion of patients whose disease had already resisted the therapy.

The trial, known as FAB-HCC, enrolled twelve patients with advanced hepatocellular carcinoma who had progressed while receiving atezolizumab plus bevacizumab. The rationale rested on a growing body of evidence that the gut microbiome can act as a rheostat for the immune system, shaping how effectively checkpoint inhibitors unleash T cells against tumors. In cancers such as melanoma, earlier studies had shown that transferring fecal material from patients who responded to immunotherapy into those who did not could sometimes rekindle the anti-tumor response. The Vienna team extended this concept in two directions: they tested the approach in liver cancer, a tumor type arising in an organ intimately connected to the gut through the portal circulation, and they compared donors of two kinds, a patient with hepatocellular carcinoma who had responded robustly to anti-PD-1 therapy and a healthy individual.

Each participant received a single fecal microbiota transplantation via colonoscopy, followed by continued treatment with atezolizumab and bevacizumab. The primary endpoint of the study was safety, measured by the incidence and severity of treatment-related adverse events. This was a critical question, because both fecal transplants and the drug combination carry theoretical risks, particularly in patients with cirrhotic livers whose immune function is already compromised. The results were reassuring. The most frequent grade 3 or 4 treatment-related adverse events were arterial hypertension, seen in five patients, and proteinuria, seen in three, both of which are well-known toxicities of bevacizumab rather than consequences of the transplant. A single case of grade 3 diarrhea was attributed to the fecal microbiota transplantation itself. Importantly, no serious treatment-related adverse events were observed, and the procedure proved feasible in this vulnerable population.

The efficacy signals, while generated in a very small cohort, were eye-catching. Four of the twelve patients, an objective response rate of 33 percent, experienced radiologically confirmed tumor shrinkage after their disease had stopped responding to the same drug combination. The median progression-free survival was 6.9 months, with a 95 percent confidence interval spanning 3.0 to 10.8 months. For a population in which rechallenge with the same regimen would ordinarily be expected to yield little benefit, these numbers suggest that modulating the gut ecosystem may genuinely change the biological terrain on which immunotherapy operates. The authors are careful to frame the trial as hypothesis-generating, and with twelve participants, chance and selection effects cannot be excluded. Still, the consistency between the clinical outcomes and the molecular findings lends the results biological plausibility.

That plausibility comes from an unusually deep set of translational analyses. By sequencing microbial DNA from stool samples collected before and after the transplant, the researchers tracked whether the donor’s microbial community took up residence in the recipient, a phenomenon known as engraftment. After the procedure, the composition of the recipients’ gut microbiomes shifted toward that of their respective donors, indicating that a single colonoscopic delivery can durably alter the intestinal ecosystem. Notably, transplants from the healthy donor appeared to engraft more successfully than those from the responding patient, an observation with practical implications for how donor material might be selected in future trials. The team also documented a meaningful ecological shift in function: bacteria previously associated with immunotherapy response became enriched after the transplant, while deleterious species waned.

The immune consequences of this microbial remodeling were visible in the blood. Following the combined intervention, the number of circulating PD-1-positive CD8-positive T cells, a population often expanded in settings of chronic antigen stimulation and immune exhaustion, decreased. At the same time, a specific subset of monocytes, characterized by the marker CD32b and high expression of CD14 with low CD16, increased. Monocytes and their derivatives are increasingly recognized as key orchestrators of the tumor microenvironment, capable of either supporting or suppressing T-cell activity, and shifts in these populations may reflect a rebalancing of innate immunity toward a more inflammation-competent state. These cellular changes provide a mechanistic bridge between the microbial intervention and the clinical responses observed on imaging.

Perhaps the most novel analytical contribution came from an antibody-sequencing approach known as phage-display immunoprecipitation sequencing, which profiles the repertoire of antibodies circulating in a patient’s blood by measuring their binding to hundreds of thousands of synthetic peptide displays. Rather than finding that overall antibody diversity or composition predicted outcome, the researchers discovered that the stability of the antibody repertoire over time was what changed following the intervention. This suggests that the humoral immune system, the arm responsible for antibody production, undergoes a measurable recalibration when the gut ecosystem is reset and checkpoint blockade is resumed. It is a reminder that cancer immunotherapy responses are systemic events, rippling far beyond the tumor itself, and that tools capable of reading these ripples may eventually help identify which patients are responding long before scans can.

The study also carries a poignant human dimension. The responding patient who served as one of the donors represents a rare example of a cancer patient directly contributing biological material that may benefit others with the same disease. The trial was conducted at a single center with support from Roche, which supplied the investigational medicinal product, and it was registered under ClinicalTrials.gov identifier NCT05750030. The authors acknowledge with gratitude the contributions of patients, donors, and their families, as well as a statistician who contributed to the trial’s design and passed away before its completion, a quiet reminder of the human effort underlying every clinical dataset.

What happens next will determine whether this approach moves from promising pilot to standard of care. The central limitation is unmistakable: twelve patients, no control group, and an open-label design in which neither patients nor physicians were blinded to the intervention. Spontaneous fluctuations in hepatocellular carcinoma, and the possibility that simply continuing atezolizumab and bevacizumab after a treatment holiday benefits some patients, cannot be ruled out. Larger, adequately powered, ideally randomized trials will be needed to confirm that fecal microbiota transplantation genuinely reverses resistance rather than merely coinciding with it. If those trials succeed, the implications would extend well beyond liver cancer. The gut microbiome is already implicated in responses to checkpoint inhibitors across melanoma, lung cancer, and renal cell carcinoma, and a safe, single-procedure intervention that reprograms it could become a universal sensitizing strategy. For now, the Vienna results offer something oncology rarely enjoys after a treatment fails: a scientifically grounded reason to try again, delivered not with a new molecule but with the microscopic residents of a healthier gut.

Subject of Research: Fecal microbiota transplantation combined with atezolizumab-bevacizumab rechallenge in immunotherapy-refractory hepatocellular carcinoma

Article Title: FMT combined with atezolizumab-bevacizumab in hepatocellular carcinoma refractory to immunotherapy: an open label, single-arm, phase II pilot study

Article References: Pomej, K., Frick, A., Scheiner, B., Baumgartner, M., Schmetterer, K. G., Reyna-Blanco, C. S., Balcar, L., Pajancic, L., Kreuter, A., Séneca, J., Hausmann, B., Köcher, T., Prinzensteiner, M., Lampichler, K., Rohr-Udilova, N., Chakrabarty, A., Klotz, A., Lampl, A. C., Zinober, K., … Pinter, M. (2026). FMT combined with atezolizumab-bevacizumab in hepatocellular carcinoma refractory to immunotherapy: an open label, single-arm, phase II pilot study. Nature Communications. https://doi.org/10.1038/s41467-026-78219-z

Image Credits: AI Generated

DOI: 10.1038/s41467-026-78219-z

Keywords: fecal microbiota transplantation, hepatocellular carcinoma, atezolizumab, bevacizumab, immunotherapy resistance, gut microbiome, checkpoint inhibitors, phase II pilot study, microbiome engraftment, PD-1, tumor immunology, Nature Communications