GLP-1s and Cancer Risk

This question arrives with more fear attached than almost any other, and it deserves a careful answer rather than either reassurance or alarm.

Start with what is established

Excess body fat is an established cause of cancer. The International Agency for Research on Cancer identifies at least 13 cancers with sufficient evidence of a causal relationship, a conclusion drawn from more than 1,000 studies: postmenopausal breast, colorectal, endometrial, kidney, liver, pancreatic, gastric cardia, gallbladder, ovarian, thyroid, oesophageal adenocarcinoma, multiple myeloma and meningioma.

That is not a hypothesis. It is decades of consistent epidemiology with plausible mechanisms — insulin resistance, chronic inflammation, and sex hormone changes driven by adipose tissue.

I lead with this because it is the comparison every discussion of hypothetical drug risk has to be measured against, and it is almost always left out. The risk from the condition is running today, quantified. The risk from the drug is, so far, theoretical.

The pancreatic cancer question

This was the first serious concern raised about the class, and it had a plausible shape: these drugs act on the pancreas, pancreatitis appears in the warnings, and chronic pancreatic inflammation is a risk factor for pancreatic cancer.

Subsequent evidence has not supported it. Large observational analyses have not shown increased incidence, and the cardiovascular outcome trials — which followed over 17,000 patients for a mean of 3.3 years in SELECT alone, counting everything that happened to them — did not produce a signal.

There is also a specific confounder worth naming. Pancreatic cancer frequently causes new-onset diabetes and unexplained weight loss before it is diagnosed. So a person can develop occult pancreatic cancer, become diabetic because of it, be prescribed a GLP-1 for the diabetes, and be diagnosed 6–24 months later. In a database, that looks like the drug preceded the cancer. It is reverse causation, and it is the single biggest trap in this literature.

Thyroid, briefly

The only cancer warning on the labels concerns thyroid C-cell tumours, derived from rodent carcinogenicity studies. Rodent and human thyroids differ in ways specifically relevant to that finding, the human data is mixed, and the absolute numbers are small on any interpretation.

It has its own article, because the detail matters and the contraindication is absolute: the thyroid cancer warning, explained.

Colorectal and the rest

Observational studies have looked at colorectal cancer with mixed and sometimes contradictory results — some suggesting reduced incidence in treated populations, others finding nothing. None of this is settled, and observational designs cannot establish causation in either direction.

Which brings me to the confounder that runs through this whole field.

Detection bias, which almost nobody adjusts for adequately

People who start these drugs see clinicians more often. They get more blood tests, more imaging, more incidental findings, and more follow-up of those findings. Looking harder finds more.

That produces an apparent increase in diagnoses that has nothing to do with biology. It is a well-recognised problem and it is difficult to fully adjust for. When you read a headline reporting increased cancer diagnoses in people taking a drug that also brings them into contact with the health system more often, this is the first thing to ask about — and it is one of the six checks in how to read a GLP-1 study.

Does the weight loss reduce risk?

Plausible, not proven for these drugs.

The closest evidence comes from bariatric surgery cohorts, which have reported reduced incidence of obesity-related cancers after substantial sustained weight loss. If the mechanism is the excess adipose tissue, then removing it should help regardless of how.

But cancer incidence is a slow endpoint. The trials have not run long enough, and nobody should tell you this is established. It is a reasonable hope with a plausible mechanism, and I would not put it more strongly than that — the honest accounting of what follow-up exists is in what we know about long-term safety.

What this means practically

Keep your cancer screening up to date — most programmes run on 1–3 year cycles and it is easy to let one slip during a year of appointments about something else. Genuinely the most useful sentence in this article. Mammography, cervical screening, colorectal screening, and skin checks according to your age and risk. Weight loss does not exempt you, and screening uptake is historically lower in people with obesity — partly because of how they have been treated in medical settings.

Do not stop screening because you feel well. Colorectal screening typically starts at 45 years and mammography in the 40s, and neither threshold moves because your weight did. Feeling well is the normal state of someone with an early cancer.

Report the things that always warranted reporting: unexplained bleeding, a new lump, a persistent change in bowel habit, unexplained weight loss beyond what the drug explains, difficulty swallowing, persistent hoarseness. That last group overlaps with the thyroid piece.

Be alert to weight loss that outpaces the drug — more than about 1.5 kg a week sustained, without a dose change to explain it. If you are losing considerably faster than the medication and your intake explain, that is worth investigating rather than celebrating.

And if you have a current or past cancer diagnosis, whether to take one of these is an oncology conversation. Not because a risk is established, but because your situation has variables an article cannot see — alongside the other situations in who should not take a GLP-1.