GLP-1s in People Being Treated for Cancer

People with both diabetes and cancer who took GLP-1 drugs had lower all-cause mortality and fewer hospitalizations than those on metformin, with a hazard ratio of 0.875.^[1] A second analysis, comparing people newly started on a GLP-1 against people newly started on insulin, produced a bigger number — and readers of this site should already be suspicious of why.

What was found

Secondary analyses found lower rates of all-cause hospitalization, sepsis, major adverse cardiovascular events, pulmonary embolism and pneumonia. Subanalyses stratified by body mass index and HbA1c did not reach statistical significance.^[1]

Which number to trust

The metformin comparison is the more believable one, and it is the smaller. Metformin is typically first-line and given to people whose diabetes is relatively well controlled — a reasonably fair comparison group.

Insulin is not. Being on insulin usually marks longer-standing, harder-to-control diabetes and more complications. Comparing a GLP-1 against insulin measures, in part, how sick each group already was before anyone counted a death.

The bigger effect came from the worse comparison. That is the pattern, not the exception.

We took this apart at length in the cancer cohorts article, where a study reported 91% lower breast cancer mortality against insulin users and nothing at all against a modern comparator in the same paper. The mechanism is identical here, and the effect size is far more modest, which is itself reassuring about this study’s quality.

The subanalyses that failed

Stratifying by BMI and HbA1c did not reach significance.^[1] That is worth noting because it removes the two most obvious explanations for how the drug might help — weight loss and glycemic control. If the benefit were driven by either, you would expect it to track them.

It may simply be that those subgroups were too small to show anything. But it means the study cannot tell you why the association exists, only that it does, in a design that cannot establish causation anyway.

What this changes for a patient

• Nothing on its own. No GLP-1 is indicated for cancer, and the authors call for prospective, well-controlled studies.
• It is reassuring about continuation. People with diabetes who develop cancer often wonder whether to stop everything metabolic. This does not suggest harm.
• The hospitalization and infection findings are relevant to cancer care specifically, where sepsis and pneumonia are major causes of treatment interruption. That is consistent with the infection study, in a much sicker population.
• It belongs with the oncologist, who is weighing appetite, nutrition and treatment tolerance in a way no general finding addresses.