Why do diabetes medicines have heart trials at all?
Regulators asked for them. Heart trials of diabetes medicines began as safety tests, required to rule out excess cardiovascular risk. The question was whether a new diabetes medicine made heart attacks or strokes more likely. Some of the trials found more than safety.
The outcome they count is called MACE, for major adverse cardiovascular events: cardiovascular death, heart attack and stroke, counted as the first of any of them to occur. SUSTAIN-6 and LEADER were designed for non-inferiority, which means they had to show the medicine was not worse than placebo by more than a set margin.
Two terms recur below. A hazard ratio (HR) compares how fast events occur in two groups. An HR of 0.80 means the treated group had events at about 80% of the comparison group's rate, a 20% lower relative rate. The range in brackets is the 95% confidence interval. If it includes 1.0, the trial could not rule out no difference at all.
What did each trial find, and in whom?
LEADER studied liraglutide in 9,340 people with type 2 diabetes at high cardiovascular risk, against placebo, for a median of 3.8 years. MACE occurred in 13.0% on liraglutide and 14.9% on placebo (HR 0.87, 0.78 to 0.97), a 13% lower relative rate. Cardiovascular death (HR 0.78, 0.66 to 0.93) and death from any cause (HR 0.85, 0.74 to 0.97) were also lower on liraglutide.
SUSTAIN-6 studied semaglutide 0.5 or 1 mg against placebo in 3,297 people with type 2 diabetes, 83% of whom had cardiovascular disease, kidney disease or both, over 104 weeks. MACE occurred in 6.6% against 8.9% (HR 0.74, 0.58 to 0.95), a 26% lower relative rate. Non-fatal stroke was less frequent on semaglutide (HR 0.61). Non-fatal heart attack did not differ significantly, and cardiovascular death was similar. Complications of diabetic retinopathy were more frequent on semaglutide (HR 1.76, 1.11 to 2.78). The trial was designed to test non-inferiority, in other words safety.
SOUL studied oral semaglutide, the medicine in Rybelsus, up to 14 mg, against placebo in 9,650 people aged 50 or over with type 2 diabetes and atherosclerotic cardiovascular disease, kidney disease or both, for a median of 49.5 months. MACE occurred in 12.0% against 13.8% (HR 0.86, 0.77 to 0.96), a 14% lower relative rate. The confirmatory secondary outcomes, including kidney outcomes, did not differ significantly.
REWIND studied dulaglutide against placebo in 9,901 people with type 2 diabetes for a median of 5.4 years. MACE occurred in 12.0% against 13.4% (HR 0.88, 0.79 to 0.99), a 12% lower relative rate. This is why dulaglutide became an active comparator in the tirzepatide trial below.
SELECT is the trial that did not require diabetes. It enrolled 17,604 people aged 45 or over with existing cardiovascular disease and a BMI of 27 or more, without diabetes. They took semaglutide 2.4 mg, the weight dose, or placebo, for a mean of 39.8 months. Cardiovascular death, heart attack or stroke occurred in 6.5% on semaglutide and 8.0% on placebo (HR 0.80, 0.72 to 0.90), a 20% lower relative rate. Everyone enrolled already had cardiovascular disease, so SELECT did not test people without it.
Is non-inferior the same as better?
No, and SURPASS-CVOT shows why. It compared tirzepatide with dulaglutide, not with placebo. It randomised 13,299 people with type 2 diabetes and atherosclerotic cardiovascular disease (13,165 were analysed) to tirzepatide up to 15 mg or dulaglutide 1.5 mg, at 640 sites in 30 countries. MACE occurred in 12.2% on tirzepatide and 13.1% on dulaglutide (HR 0.92, 95.3% confidence interval 0.83 to 1.01).
That result met the trial's test for non-inferiority (P=0.003). Tirzepatide was shown to be at least as good as dulaglutide. It was not shown to be better: the test for superiority gave P=0.09, and the US label states that superiority was not established. Both tests turned on the upper end of the confidence interval. Below 1.05 counted as non-inferior. Below 1.00 would have shown that tirzepatide was better. It reached 1.01, so a difference of zero cannot be ruled out. Because dulaglutide has heart evidence of its own, from REWIND, the trial does not show how tirzepatide compares with placebo.
Participants were followed for a median of 210.1 weeks, about four years, and the trial as a whole ran about five years. Digestive events were more frequent on tirzepatide.
Death from any cause was 8.5% on tirzepatide and 10.1% on dulaglutide (HR 0.84, 0.75 to 0.94). That is a secondary finding the trial was not designed to confirm, because its statistical plan did not control for testing several outcomes. It is not a proven result. A later exploratory analysis of a six-part heart and kidney composite found 23.7% against 27.4% (HR 0.84, 0.79 to 0.90), with digestive events in 42.5% against 35.9%. Exploratory analyses raise questions. They do not settle them.
What does the headline number leave out?
Relative and absolute figures answer different questions. In SELECT, a 20% lower relative rate (HR 0.80) was 6.5% against 8.0% over a mean of 39.8 months. That is about 15 fewer events per 1,000 people, from 80 to 65 per 1,000. A relative figure standing alone sounds larger than the absolute change.
Background care matters. In SURPASS-CVOT, 86% of participants were taking statins, 83% antiplatelet medicines and 65% beta blockers, and 9% were Asian. The results describe people who were already receiving that care.
Side effects carry a cost. In SELECT, 16.6% of people on semaglutide 2.4 mg stopped treatment because of adverse events, against 8.2% on placebo. In SURPASS-CVOT there were more digestive events on tirzepatide. Those effects are set out in the article on weight-loss injection side effects.
Which brand carries which heart use, and where?
It depends on the country and the brand. In Malaysia, on the NPRA register read on 30 September 2026, Ozempic (semaglutide, a weekly injection) is registered for type 2 diabetes, and its package insert of 19 November 2025 adds a use of reducing MACE in adults with type 2 diabetes who have established cardiovascular disease or are at high cardiovascular risk. Rybelsus (semaglutide tablets) is registered for glycaemic control in type 2 diabetes only. Wegovy (semaglutide 2.4 mg) is registered for weight management in adults and has no heart indication. Mounjaro (tirzepatide) is registered for type 2 diabetes and weight management in adults and has no heart indication. Saxenda (liraglutide) is registered for weight management in adults and in adolescents from 12 years with obesity and a body weight above 60 kg. The medicines themselves, their doses and their weight trials are compared in how the GLP-1 medicines differ.
In the United States the labels differ. Ozempic covers type 2 diabetes, MACE in type 2 diabetes with established cardiovascular disease, and kidney outcomes in type 2 diabetes with chronic kidney disease. Rybelsus covers type 2 diabetes and MACE in type 2 diabetes at high risk. The Wegovy injection covers weight management and MACE in adults with established cardiovascular disease and obesity or overweight. Mounjaro covers type 2 diabetes and, since 28 August 2026, MACE in adults with type 2 diabetes at high risk. Zepbound (tirzepatide), the US weight brand, has no heart indication, and it was not on the Malaysian register when searched on 30 September 2026.
A heart approval means a regulator reviewed the trial evidence and judged that, for the group of people described on the label, the benefits of that heart use outweigh its risks. It is a decision about one product, one population and one country. It does not say the medicine helps everyone who takes it, or that it is better than other treatments. It does not extend to people outside the group described, or to another brand of the same medicine that does not carry the use. And it does not cross borders: Malaysia's register makes its own decisions, as the difference above shows. Whether an approval applies to a given person is a medical decision, made with that person's own doctor.
What did the heart failure and kidney trials show?
These trials are narrower. SUMMIT studied tirzepatide against placebo in 731 people with heart failure with preserved ejection fraction (an ejection fraction of 50% or more) and a BMI of 30 or more, for a median of 104 weeks. Cardiovascular death or worsening heart failure occurred in 9.9% against 15.3% (HR 0.62, 0.41 to 0.95), a 38% lower relative rate. That composite was driven by worsening heart failure events (8.0% against 14.2%, HR 0.54). Cardiovascular death itself was 2.2% against 1.4% (HR 1.58, 0.52 to 4.83), a range too wide to separate a rise from a fall. SUMMIT did not show a fall in cardiovascular death. A symptom and function score rose 19.5 points against 12.7, and 6.3% against 1.4% stopped because of adverse events.
STEP-HFpEF studied semaglutide 2.4 mg in 529 people for 52 weeks. A symptom score rose 16.6 points against 8.7 on placebo, weight fell 13.3% against 2.6%, and the six-minute walk lengthened by 21.5 m against 1.2 m. It measured symptoms and function. It was not designed to count heart events.
FLOW studied semaglutide 1.0 mg against placebo in 3,533 people with type 2 diabetes and chronic kidney disease, for a median of 3.4 years, and was stopped early for efficacy. The kidney composite had an HR of 0.76 (0.66 to 0.88), a 24% lower relative rate. Cardiovascular death (HR 0.71), MACE (0.82) and death from any cause (0.80) were also lower on semaglutide. The decline in kidney filtration rate (eGFR) was slower by 1.16 ml/min/1.73 m2 a year. The result applies to people with both conditions, whose kidney care sits with their own doctor.
What does a heart approval not mean for someone taking these for weight?
Most people who ask about these medicines are asking about weight, and the heart results above do not automatically apply to them. SELECT enrolled only people with existing cardiovascular disease. SURPASS-CVOT enrolled only people with type 2 diabetes and atherosclerotic disease. For tirzepatide in people with obesity and no diabetes, the outcome trial, SURMOUNT-MMO, is active and no longer recruiting, with primary completion estimated for October 2027. As of 30 September 2026 it has no result.
What happens after stopping matters too. The extension of the STEP 1 trial analysed 327 people, from both the semaglutide and the placebo groups, over the year after treatment stopped. Those who had taken semaglutide 2.4 mg had lost an average of 17.3% of their weight by week 68, and by week 120 they had regained an average of 11.6 percentage points of it. In that group, most measured heart and metabolic risk factors moved back towards their starting values. No trial has yet measured a heart outcome after stopping. The weight side of this is covered in rebound weight after stopping GLP-1 medication.
These are prescription medicines. In Malaysia they are prescribed by a registered doctor and supplied by the doctor or a licensed pharmacist, after a full evaluation: a history, an examination and usually blood tests. What a doctor checks before prescribing GLP-1 medication sets out that assessment, and a weight management follow-up what is reviewed afterwards.
Nothing in this article is a reason to ask for one of these medicines, or to start, switch or stop any medicine. Heart, kidney and diabetes treatment decisions belong with the person's own doctor, who knows the full history and every medicine being taken.
