At a therapeutic dose, some cancers, against some comparators. A retrospective network study set 662,013 nondiabetic GLP-1 users against 272,343 bariatric surgery patients and 158,446 users of other weight-loss drugs. At therapeutic dose, GLP-1 use was associated with lower colorectal cancer incidence than surgery, HR 0.72, and lower pancreatic cancer incidence, HR 0.63. Liver cancer did not move.
The dose qualifier carries the whole result. At any dose rather than therapeutic dose, none of the differences reached significance. A microdose is not what was studied here.
One study in a narrower population reports a mortality gap without the mechanism behind it. Among men with prostate cancer on androgen deprivation therapy, GLP-1 use was linked to lower all-cause mortality at HR 0.60, while cardiac events did not differ [2]. That is a gap with nothing cardiac behind it, despite cardiovascular harm being the stated reason for studying the group at all.
Weight loss after a cancer diagnosis is a different question again, and a symptom before it is a goal[3].
This desk prices bariatric surgery against a year of medication, most recently in what you pay against what surgery costs. This study puts a different item on that same comparison — cancer incidence — and finds the drug arm ahead on two of three cancers [1].
Among nondiabetic adults with obesity, therapeutic-dose GLP-1 use was associated with a lower incidence of colorectal cancer than bariatric surgery, at a hazard ratio of 0.72, and of pancreatic cancer at 0.63. Hepatocellular carcinoma showed no significant difference. Against other weight-loss medications — orlistat, phentermine with topiramate, naltrexone with bupropion — colorectal cancer came in at 0.68 with no significant difference on the other two. Those older drugs are the ones this desk rarely has to price, because almost nobody asks for them; the money goes where one build carries eight prices.
Two things about the statistics belong in any summary. The published abstract gives hazard ratios and P values with no confidence intervals, so the precision of these estimates is not available here and this desk will not estimate it. And the two P values against surgery, .038 and .041, sit just under the conventional threshold across a set of comparisons covering three cancers, two comparator groups and two dose strata.
On the comparison itself, the surgery arm is the harder one to interpret. People who undergo bariatric surgery differ from people prescribed a drug in ways propensity matching addresses only partly, and surgery itself alters gastrointestinal anatomy in ways that plausibly affect cancer surveillance and detection. Nothing in a claims network settles that.
For a buyer, the honest framing is that no seller can price this. A reduced cancer hazard over years is not a line item on a monthly quote, and the evidence here is a retrospective association at one dose stratum in one network. This desk records it as a reason the drug-versus-surgery comparison is broader than the price columns in the comparison surgery won suggest — not as a benefit anyone should pay a premium for today.