- Design
- systematic review and meta-analysis of randomised controlled trials, searched to 17 March 2025
- Population
- 48 randomised trials, 48,765 patients with type 2 diabetes, SGLT2 inhibitor versus placebo or active comparator
- Primary outcome
- gastrointestinal neoplasms, mostly captured as reported adverse events rather than adjudicated cancer endpoints
- Effect
- OR 1.10 (95% CI 0.84-1.44), P=0.46, I-squared 0%; all site-specific analyses non-significant
Concern about carcinogenicity has followed SGLT2 inhibitors since their approval, driven largely by imbalances in small numbers of events in individual trials. This meta-analysis pooled 48 randomised trials and 48,765 patients with type 2 diabetes comparing an SGLT2 inhibitor against placebo or an active comparator, taking gastrointestinal neoplasms from publications, supplementary material and trial registries.
There was no association overall: odds ratio 1.10 (95% CI 0.84 to 1.44), P=0.46, with I-squared of 0%. Site-specific analyses were all non-significant, including oesophageal 1.12 (0.37 to 3.45), gastric 1.20 (0.65 to 2.23), hepatic 0.62 (0.31 to 1.22), pancreatic 0.91 (0.51 to 1.64), colonic 1.28 (0.78 to 2.08), colorectal 0.76 (0.27 to 2.17) and rectal 0.98 (0.49 to 1.97). Subgroup analyses by agent, age, body mass index, HbA1c, dose and treatment duration were also negative.
The authors are careful about what this can and cannot support, and the caveats are structural rather than statistical. Around half the trials followed patients for a year or less, which is far shorter than the latency of a solid tumour. Cancers were captured as reported adverse events rather than as centrally adjudicated endpoints, so ascertainment was neither systematic nor blinded to treatment. Event counts at individual sites were small, which is why intervals like 0.37 to 3.45 for oesophageal cancer are compatible with almost anything.
The usable conclusion is that there is no short-term signal to act on. A gastroenterologist asked whether a patient with Barrett's oesophagus or a history of colonic adenomas should avoid an SGLT2 inhibitor can say that randomised evidence gives no reason to, while noting that long-term oncological safety has not been established by trials of this length.
- Do not withhold an SGLT2 inhibitor over gastrointestinal cancer risk on current evidence
- Say plainly that follow-up in these trials is too short to address long-term risk
- Keep existing surveillance decisions, for Barrett's oesophagus or previous adenomas, driven by the underlying condition rather than by diabetes therapy
- Treat wide site-specific intervals as absence of data, not as evidence of safety at that site
- Report suspected drug-related neoplasms through pharmacovigilance, since trial adverse-event capture is what these pooled figures rest on
The statistics, in plain English
An I-squared of 0% means the trials agreed closely with each other, so the pooled estimate is a fair summary rather than an average of conflicting results. The overall interval, 0.84 to 1.44, still allows a 44% relative increase, so this is not proof of no effect; it is failure to find one. The site-specific intervals are much wider because each rests on a handful of cancers, and an interval from 0.37 to 3.45 is uninformative in both directions. With half the trials running a year or less and cancers recorded as adverse events rather than adjudicated, absence of a signal here is weaker evidence than the same result from a long-term cancer registry would be.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for gastroenterology & hepatology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free