In 2019, researchers launched JAVELIN Head and Neck 100, a major clinical trial testing whether adding the immunotherapy drug avelumab to standard chemoradiotherapy could improve survival in patients with locally advanced head and neck cancer. The trial enrolled 697 patients across 22 countries, but the results, announced two years later, were disappointing: avelumab failed to improve progression-free survival or overall survival. The drug appeared to make no difference at all.
That failure puzzled researchers. Immunotherapy works, sometimes brilliantly, in this cancer type. So why did this trial miss? A new analysis by Haas et al. published this month in Nature Cancer offers a surprising answer: bacteria living inside the tumors were making patients resistant to immunotherapy, and no one was looking for it.
The finding in plain terms
Researchers led by teams at Cleveland Clinic, Memorial Sloan Kettering, and the University of Chicago re-analyzed tumor samples from the original JAVELIN trial using genomic and microbiome sequencing. Their finding was striking: patients whose tumors harbored high levels of bacteria had substantially worse outcomes with avelumab than patients whose tumors were bacteria-poor.
This matters because bacteria inside solid tumors are not normal contamination; they actively reshape the immune microenvironment. When bacteria colonize a tumor, they trigger an influx of immune cells called neutrophils. While neutrophils are essential for fighting infections, inside a tumor they suppress T cells, the immune cells that checkpoint inhibitors like avelumab are designed to activate. In effect, bacteria-laden tumors silence the very immune system that immunotherapy tries to wake up.
Why it matters
Head and neck cancer immunotherapy has been one of oncology’s bright spots. Checkpoint inhibitors targeting PD-L1 and PD-1 have improved survival in patients with advanced disease. But immunotherapy doesn’t work for everyone, and the field lacks reliable predictors of who will respond. Standard biomarkers, such as tumor PD-L1 expression, are imperfect; some patients with high PD-L1 tumors still fail treatment.
This study identifies a concrete, mechanistic explanation for some of that failure. If intratumoral bacteria drive immunotherapy resistance in a subset of patients, then antibiotics that reduce bacterial burden might restore immunotherapy sensitivity. That’s not speculative; the researchers showed in mouse models that antibiotic treatment lowered neutrophil infiltration, expanded T cell populations, and enhanced tumor control.
For the field, this also explains the JAVELIN Head and Neck 100 failure. Avelumab is an effective anti-PD-L1 drug in other contexts. The failure wasn’t because the drug didn’t work; it was because a substantial fraction of patients had a biological barrier to response that no one had identified.
How they did it
The study analyzed tumor tissue and blood samples from 355 patients in the JAVELIN trial who had material available for analysis. Researchers performed whole-genome sequencing to identify genetic mutations, immune profiling to map immune cell populations, and 16S ribosomal RNA sequencing to quantify and identify bacteria colonizing the tumors.
They then asked: which microbiome, immune, and genetic features predict response to avelumab plus chemoradiotherapy versus chemoradiotherapy alone?
The association between high intratumoral bacteria and poor avelumab response was robust and reproducible. When researchers divided patients into quartiles by bacterial burden, those in the highest quartile (most bacteria) had dramatically worse outcomes. Mechanistically, high-bacteria tumors showed elevated neutrophil infiltration, depleted T cell populations, and markers of immune suppression; this profile was distinct from low-bacteria tumors.
To test causation, researchers inoculated mouse tumors with bacterial species isolated from patient samples, then treated mice with or without antibiotics. Antibiotic treatment reduced bacterial load, decreased neutrophil accumulation, restored CD8 T cell populations, and significantly slowed tumor growth.
Limitations and caveats
Several important limitations shape how we should interpret this finding:
This is post-hoc analysis of a failed trial. The JAVELIN trial wasn’t designed to measure intratumoral bacteria. The microbiome analysis was performed retrospectively on available samples, which may not be fully representative of the original trial cohort. Post-hoc subgroup analyses, even if biologically plausible, require prospective validation.
Sample availability was incomplete. Of 697 trial patients, microbiome data were only available for 355 (51%). If tumor sampling or processing was not random, this could introduce bias.
The bacteria-immunotherapy link is shown in mouse models, not yet in patients. The antibiotic experiments were done in orthotopic murine tumors. While mouse models are useful for testing mechanism, mice are not humans. Intratumoral bacteria exist in human head and neck cancers, but whether treating bacteria actually restores immunotherapy response in patients remains unknown.
Causation is inferred, not proven. The correlation between bacterial burden and immunotherapy resistance is clear. The mechanism (neutrophil suppression of T cells) is plausible and supported by cell-type profiling. But the data are correlative and mechanistic, not a direct causal experiment in humans.
Small subgroup with ultra-high bacteria. The most extreme bacteria-rich tumors were rare; only a minority of patients had the very highest bacterial loads. This means the effect size, while impressive in that group, may not apply to most patients.
What this means in practice
For clinicians treating head and neck cancer now, this finding is not yet actionable. No clinic is measuring intratumoral bacterial burden, and the data arguing for antibiotic co-treatment remain preclinical.
However, this work has triggered real clinical investigation. The American Cancer Society is currently funding a trial testing whether antibiotics can modulate the intratumoral microbiome and improve immunotherapy response in patients with head and neck squamous cell carcinoma. Results from that trial, expected in the coming years, will determine whether this insight translates to the clinic.
For researchers, the implication is clear: the intratumoral microbiome is not a detail but rather a biologically important component of the tumor ecosystem. This probably applies beyond head and neck cancer. Other tumor types also harbor intratumoral bacteria. Understanding how bacterial communities suppress immunity in colorectal, lung, and pancreatic cancers may unlock resistance mechanisms in those settings too.
For patients enrolled in immunotherapy trials, this underscores why comprehensive characterization of tumor biology matters, not just mutation status or PD-L1 staining. The bacteria are invisible on an imaging study or in a routine pathology report, yet they shape treatment outcomes.
Source and further reading
Primary paper: Haas et al., “Tumor ecosystem and microbiome features associated with efficacy and resistance to avelumab plus chemoradiotherapy in head and neck cancer,” Nature Cancer 7, 98–115 (2026). https://doi.org/10.1038/s43018-025-01068-0
Related papers from the same research consortium published simultaneously (January 2026):
- Rieth et al., “Intratumoral bacteria are immunosuppressive and promote immunotherapy resistance in head and neck squamous cell carcinoma,” Nature Cancer (2026). https://doi.org/10.1038/s43018-025-01067-1
- Geller et al., “Tumor bacterial burden dictates immunotherapy fate in head and neck cancer,” Nature Cancer (2026). https://doi.org/10.1038/s43018-025-01074-2
These three papers form a cohesive story: intratumoral bacteria suppress T cell immunity, high bacterial burden predicts immunotherapy resistance, and antibiotic treatment restores immune function in preclinical models.
Original trial reference: Ferris et al., “Primary results of the phase III JAVELIN head & neck 100 trial: Avelumab plus chemoradiotherapy (CRT) followed by avelumab maintenance vs CRT in patients with locally advanced squamous cell carcinoma of the head and neck (LA SCCHN),” Annals of Oncology 31 (2020): S1215. https://doi.org/10.1016/j.annonc.2020.10.472