Before semaglutide was a weight-loss story, it was a diabetes drug that had to prove it would not harm the heart. SUSTAIN-6 was that proof — a regulatory safety trial that, almost as a byproduct, delivered the first randomized signal that semaglutide might actually reduce cardiovascular events. It is a trial worth reading carefully, because it also carries the most honest caveat in the semaglutide file: a real, statistically significant increase in diabetic-retinopathy complications.
Why it mattered: a safety trial that surprised everyone
After 2008, U.S. regulators required new type 2 diabetes drugs to demonstrate they did not raise cardiovascular risk. SUSTAIN-6 was designed to meet that bar — a non-inferioritytrial whose primary job was to rule out harm, with a pre-specified margin.[1] What made it notable is that it did not merely clear the safety bar; the same data crossed into superiority, hinting that semaglutide lowered cardiovascular events rather than simply leaving them unchanged. That distinction — designed to show “no worse,” but reading out as “better” — is the crux of how SUSTAIN-6 is often over-read, and why its framing deserves care.
The design
SUSTAIN-6 randomized 3,297 patients with type 2 diabetes at high cardiovascular risk to once-weekly subcutaneous semaglutide (0.5 mg or 1.0 mg) or matching placebo, on top of standard care, over 104 weeks (about two years).[1] The primary outcome was the first occurrence of a three-point composite — death from cardiovascular causes, nonfatal myocardial infarction, or nonfatal stroke (MACE). Crucially, this was an event-driven trial powered to test non-inferiority for that composite, not a study built and powered from the outset to demonstrate a reduction. It sits alongside the later obesity program described in the semaglutide weight-loss trials, but its population and intent were different: cardiovascular safety in diabetes.
The primary result
First MACE occurred in 6.6% of the semaglutide group versus 8.9%of the placebo group — a hazard ratio of 0.74 (95% CI 0.58–0.95), which met non-inferiority and also reached statistical significance for superiority.[1] Looking under the composite, the benefit was uneven: nonfatal stroke fell meaningfully (HR 0.61), nonfatal myocardial infarction trended lower without individual significance (HR 0.74), and cardiovascular death was essentially unchanged (HR 0.98).[1] In other words, the headline number was driven largely by reduced stroke rather than by a uniform effect across every component — a nuance that matters when the single figure “0.74” gets quoted on its own.
The retinopathy signal — the honest caveat
SUSTAIN-6's most important complication was not cardiovascular. Complications of diabetic retinopathy — vitreous hemorrhage, blindness, or the need for treatment with an intravitreal agent or photocoagulation — were significantly more common on semaglutide: 3.0% versus 1.8% on placebo, a hazard ratio of 1.76 (95% CI 1.11–2.76).[1] This was not a chance blip to be waved away; it was a pre-specified adverse-event category that crossed significance.
A later post-hoc analysis put the finding in context: the excess was concentrated in patients who already had retinopathy and who were on insulin at baseline, and it tracked with the magnitude and speed of the early HbA1c reduction — consistent with the long-recognized phenomenon of “early worsening” of retinopathy when glucose is lowered rapidly, rather than a direct toxic effect of the drug on the retina.[2] That is a more reassuring mechanism, but it does not erase the signal, and it remains a genuine consideration for people with pre-existing eye disease. It is the kind of caveat that a fair reading of this trial has to keep in view.
Limitations
Several constraints bound how far SUSTAIN-6 can be pushed. It was designed for non-inferiority, not powered from the start to prove superiority, so the superiority finding — while formally significant — should be read as hypothesis-strengthening rather than definitive. It was event-driven and relatively short at about two years, with a modest absolute number of events. The retinopathy signal is a real caveat. And the population was diabetes-only and high-risk, so the result does not automatically transfer to people with obesity but without diabetes. Cross-trial comparisons — including with the dual agonist covered in tirzepatide vs semaglutide evidence — should be made cautiously, given different populations, endpoints, and durations.
How it set up SELECT
SUSTAIN-6 left an obvious open question: if semaglutide cut cardiovascular events in people with diabetes, was the benefit tied to glucose control, or to something broader — weight, inflammation, vascular biology — that would apply even without diabetes? That question is precisely what the much larger SELECT trial was built to answer, in adults with established cardiovascular disease and obesity but without diabetes. SUSTAIN-6 was the smaller, earlier signal; SELECT was the dedicated test. Read together, they trace how a mandated diabetes safety study became the seed of a cardiometabolic outcomes program.
This article summarizes published trial results and is for general education, not medical advice. Semaglutide's cardiovascular benefit in SUSTAIN-6 came with a real retinopathy signal; decisions about GLP-1 therapy — especially with pre-existing eye disease — belong with a qualified clinician who knows your history.