Antibiotics and Immunotherapy: Can the Same Drug Help and Harm Cancer Treatment?

Antibiotics and Immunotherapy: Can the Same Drug Help and Harm Cancer Treatment?

Antibiotics are now firmly linked to worse outcomes in patients on checkpoint inhibitors, the immunotherapy drugs that release the brakes on the immune system so it can attack cancer. But in bowel cancer, some bacteria live inside the tumour and protect it from the immune system. Killing those bacteria helps. One drug. Two places in the body. Opposite results. What matters may be which antibiotic is given and where in the body it works.

Start with the part that is hardest to dispute. A meta-analysis by Porcari and colleagues in Cancer Cell pooled data from 116 studies involving 97,850 patients on immunotherapy. Those who took antibiotics had about a 49% higher risk of dying and a 44% higher risk of their cancer growing. The practical message is to use antibiotics carefully: treat real infections, choose the drug that kills the fewest other bacteria, and stop as soon as it is safe.

That message rests on a simple idea: bacteria help immunotherapy, so killing them hurts it. Another line of research points the other way. In some tumours, bacteria sit inside the tissue and hold the immune system back, and killing them appears to help. So if removing the wrong bacteria harms immunotherapy, could removing the right ones help it?

Antibiotics

Antibiotics and Immunotherapy: What a 97,850-Patient Meta-Analysis Reveals

One Word, Many Different Drugs

Almost every study asks one thing: did the patient take an antibiotic, yes or no? Imagine two patients. One took a few tablets for a sore throat. The other spent two weeks in hospital on a strong antibiotic through a drip. Both are counted the same way.

For the bacteria in the gut, they are not the same at all. Broad-spectrum antibiotics kill many species at once, including the ones linked to better immunotherapy results. Two studies published in Science in 2018, by Gopalakrishnan and colleagues and by Routy and colleagues, found that patients who did well carried more of these helpful bacteria, especially Faecalibacterium prausnitzii. It makes butyrate, a fatty acid that feeds the gut lining and calms inflammation. A short course of a narrow-spectrum antibiotic leaves most of that community alone.

The Cancer Cell review admits the gap. The pooled data could not separate the drug class from the dose or the length of the course because studies reported these details so inconsistently. So one average number warns doctors without telling them which antibiotic to choose.

When Bacteria Live Inside the Tumour

Most of this discussion is about the gut. But bacteria also live inside tumours, and there the story reverses.

Bullman and colleagues showed in Science in 2017 that a bacterium called Fusobacterium nucleatum lives inside bowel tumours and travels with the cancer when it spreads to the liver. They grew these human tumours in mice and gave the mice metronidazole. Bacterial levels fell, and the tumours grew less. Jiang and colleagues, writing in Cell Host & Microbe in 2023, explained why it matters. Patients whose cancer resisted immunotherapy carried more F. nucleatum, which releases a chemical called succinic acid. The chemical switches off the tumour’s alarm signal, so the killer T cells never come in. In mice, metronidazole reversed all of it.

The pattern is not limited to the bowel. In head and neck cancer, Silver and colleagues, writing in Nature Cancer, found that the number of bacteria inside a tumour predicted response to immunotherapy better than the two tests doctors normally use. Here, the bacteria hold the immune system back, and removing them becomes the treatment.

A Drug That Opens Only Inside the Tumour

The two lines of evidence now meet. Metronidazole kills F. nucleatum, but it is also an antibiotic that kills a range of anaerobic bacteria. Swallowed or given through a drip, it travels everywhere, potentially helping inside the tumour and harming beneficial bacteria in the gut. Researchers hunting for a drug that kills F. nucleatum and nothing else call this their biggest problem.

One answer is to change the delivery rather than the drug. Wang and colleagues, writing in Nature Biotechnology in 2024, packed an antibacterial compound into tiny fat bubbles called liposomes. The bubbles stay shut as they travel and open only in the acid inside a tumour, so the gut is left alone. More than 70% of infected mice survived long term. Killing the bacteria also left fragments behind that trained the immune system, which then attacked tumours carrying no bacteria at all.

The same paper reported that patients having bowel cancer surgery stayed free of disease for longer if they had been given antibiotics from the family metronidazole belongs to or the one that includes clindamycin. That finding comes from old records, so it cannot establish cause and effect. But the question has changed. It is no longer whether to give an antibiotic. It is where it should act.

What Would It Take to Test This?

The Cancer Cell authors want future studies to record which antibiotic was given, why it was given, at what dose, by which route, and for how long. They also want researchers to collect repeated samples of the patient’s gut bacteria. That list would answer most of the open questions about the gut.

My question needs one more thing: samples from the tumour as well as the stool. Nearly all the evidence so far comes from stool, which reflects only the gut microbiome, or from prescription records with no bacterial data at all. A study sampling both places could ask whether killing gut bacteria harms the patient while killing tumour bacteria helps, and whether the bacteria in a tumour mark out the patients a targeted drug would suit. Until then, both effects stay hidden inside one average number.

What This Means Now

Nothing here changes what should happen in clinic tomorrow. Antibiotics remain essential in cancer care, and the targeted methods have only been tested in mice. Careful use stays the safe advice. But the biology points further than avoidance. The gut and the tumor are two different worlds of bacteria, and the same drug can do harm in one and good in the other. If that holds, the goal is to choose and deliver these drugs with far more precision.

Written by Dr Mazen Al-Obaidi.

 References

  1. Porcari S, et al. Cancer Cell. 2026;44(11):2132–2153.
    doi:10.1016/j.ccell.2026.09.012
  2. Bullman S, et al. Science. 2017;358(6369):1443–1448.
    doi:10.1126/science.aal5240
  3. Jiang SS, et al. Cell Host Microbe. 2023;31(5):781–797.e9.
    doi:10.1016/j.chom.2023.04.010
  4. Wang M, et al. Nat Biotechnol. 2024;42(8):1263–1274.
    doi:10.1038/s41587-023-01957-8
  5. LaCourse KD, et al. PMC10790632. Cell Rep. 2022;41(7):111625. doi:10.1016/j.celrep.2022.111625
  6. Silver NL, et al. Nat Cancer. 2026;7(1):80.
    doi:10.1038/s43018-025-01067-1, PMC12823012
  7. Gopalakrishnan V, et al. Science. 2018;359(6371):97–103.
    doi:10.1126/science.aan4236
  8. Routy B, et al. Science. 2018;359(6371):91–97.
    doi:10.1126/science.aan3706

FAQ

What is immunotherapy?

The type of immunotherapy discussed here uses drugs that take the brakes off the immune system so the patient's own T cells can attack the tumour.

Why would an antibiotic interfere with that?

The immune system is shaped by the bacteria living in the gut. Antibiotics kill those bacteria. When the helpful ones go, the immune attack on the tumour appears to weaken.

How strong is that link?

Strong. One study combined the results of many others, covering almost 98,000 patients. Those who took antibiotics were more likely to die and more likely to see their cancer grow. The same pattern showed up in many types of cancer and in many countries.

Does that prove antibiotics caused the worse outcomes?

No. Patients who need antibiotics are often sicker to begin with, and sicker patients do worse regardless. The studies also never measured anyone's bacteria, so the proposed mechanism was never checked directly.

Should a cancer patient refuse antibiotics, then?

No. An untreated infection can quickly become life-threatening. The message is to treat real infections, use the narrowest drug that works, and stop as soon as it is safe.

What is the twist?

Bacteria do not only live in the gut. Some live inside the tumour itself. There, they shut down the immune system's alarm signals and keep killer T cells out. In those tumours, killing bacteria helps rather than harms.

Which bacterium is this?

Fusobacterium nucleatum, mainly. It is found inside colorectal tumours and travels with them when the cancer spreads to the liver. In head and neck cancer, the total amount of bacteria in a tumour predicts immunotherapy failure better than the standard markers do.

Why not simply give an antibiotic that kills it?

Because the drug that kills it, metronidazole, also kills the helpful bacteria in the gut. Taken as a tablet or given through a drip, it travels everywhere in the body, so it reaches the gut and the tumour at the same time. It helps in one place and harms in the other. That is the whole problem.

Is any of this available to patients?

No. The targeted work is in mice. The supporting human data come from look-back studies, which cannot prove cause. This is a research direction, not a treatment.

What would settle the question?

A study that collects both stool and tumour samples from the same patient. Every study so far looks at one or the other, so the two opposite effects stay blended into a single number.

What is the practical takeaway?

Keep using antibiotics carefully. But stop treating “antibiotics” as one thing. Which drug is given, how long it is given for, and where it acts in the body may matter more than whether one was given at all.