The Heart Data Nobody Brings Up at Dinner

Most of the public conversation about these drugs is about how people look. Almost none of it is about the trial that, in my view, mattered most — and it is the one I lead with when I am writing a letter of medical necessity to an insurer.

What SELECT did

SELECT enrolled more than 17,000 adults who had established cardiovascular disease — a prior heart attack, stroke, or peripheral arterial disease — and a BMI of 27 or above, and who did not have diabetes. That last exclusion is the whole point. Every prior cardiovascular outcomes trial of a GLP-1 had been run in people with type 2 diabetes, where you could always argue the benefit came from glucose control.

Participants received weekly semaglutide 2.4 mg or placebo, on top of whatever guideline-directed cardiovascular care they were already receiving — statins, antiplatelets, antihypertensives, the standard armamentarium. Median follow-up was over three years.

The primary endpoint was a composite of cardiovascular death, non-fatal myocardial infarction, and non-fatal stroke.

The result: about a 20% relative reduction in that composite. Hazard ratio 0.80. Statistically robust, and in a population already on good background therapy.

Why the timing of the curves is interesting

Here is the detail that made cardiologists sit up.

If a drug prevents heart attacks purely because it makes people smaller, you would expect the benefit to appear late — after enough weight has come off to shift blood pressure, lipids, and glycemia, and then after enough time for that improved risk profile to translate into fewer events. Years, plausibly.

The event curves in SELECT began to separate earlier than that reasoning comfortably accommodates. Not dramatically early, and reasonable people read the Kaplan-Meier plots differently. But early enough that “it’s just the weight loss” became a harder position to hold without qualification.

The competing explanation is that GLP-1 receptor agonism has direct effects on the vasculature and on inflammation. GLP-1 receptors are present in the heart, in vascular endothelium, and on immune cells. These drugs reduce C-reactive protein substantially, and they appear to have effects on endothelial function and possibly on plaque biology that are not simply downstream of body weight.

I want to be careful: this is a mechanistic hypothesis with supporting evidence, not a settled fact. SELECT was not designed to disentangle weight-mediated from weight-independent effects, and formal mediation analyses of trials like this are notoriously slippery. What I will say is that the “it’s only the weight” reading is no longer the default, and among the cardiologists I refer to, most have stopped assuming it.

The clinical implication is the same either way. The mechanistic question is about what we call this class of drug — and calling it “cardiovascular therapy” changes who gets offered it.

The neighbouring evidence

SELECT does not sit alone.

SUSTAIN-6 had shown, back in 2016, that semaglutide reduced cardiovascular events in people with type 2 diabetes and high cardiovascular risk. That was the first signal.

STEP-HFpEF studied semaglutide in people with heart failure with preserved ejection fraction and obesity — a condition with almost no effective drug therapy for decades — and found meaningful improvements in symptoms, physical limitation, and exercise capacity. For a group of patients who had been repeatedly told there was nothing to offer them, that was a substantial result.

Kidney outcomes have moved in the same direction, with trial evidence supporting benefit on progression of chronic kidney disease in people with type 2 diabetes.

Taken together, this looks less like a weight loss drug that happens to have side benefits and more like a metabolic and cardiovascular drug whose most visible effect is weight.

The caveat I insist on

SELECT studied people with established cardiovascular disease. Prior heart attack, prior stroke, symptomatic peripheral arterial disease. That is a population with a high baseline event rate, which is precisely why a 20% relative reduction translates into a meaningful absolute number of prevented events.

You cannot lift that 20% and apply it to a healthy 41-year-old with a BMI of 29 and no cardiac history. Her absolute risk over three years is low, so the same relative reduction would prevent very few events. Relative risk reduction without absolute risk is one of the most reliable ways to mislead people, including yourself, and it happens constantly in health writing.

Whether these drugs prevent first cardiovascular events in lower-risk populations is a genuinely open question, and it would take a much larger and longer trial to answer.

Why I bring it up anyway

Two practical reasons.

Coverage. When an insurer denies a prescription as cosmetic, the outcomes data is the argument. Not “my patient wants to lose weight” but “my patient has established cardiovascular disease and there is a large randomised trial showing a 20% reduction in major adverse cardiovascular events in exactly this population.” Different letter, different result, in my experience more often than not.

Other outcome data has followed the same pattern. FLOW reported a 24% reduction in major kidney events in type 2 diabetes with chronic kidney disease, and SURMOUNT-OSA led to the first drug indication for obstructive sleep apnea — covered in GLP-1s and kidney disease and GLP-1s and sleep apnea. Each strengthens the same argument, and each is a specific coded indication rather than a general one.

The trial list has kept growing. STEP-HFpEF and SUMMIT in heart failure with preserved ejection fraction, STEP 9 in knee osteoarthritis, and randomised liver trials in steatohepatitis — each a specific coded indication with evidence behind it, covered in heart failure, knee osteoarthritis and fatty liver disease.

Read the absolute numbers, not just the relative ones. A 20% relative reduction is not the same claim as a 20 percentage point one, and the difference decides what it means for you — how to read a GLP-1 study.

Framing. Patients who understand that they are taking a cardiovascular drug relate to it differently than patients who understand they are taking a diet drug. They are less likely to stop it the moment they hit a goal weight. They ask better questions. They are, frankly, less ashamed — and shame is a major reason people quit medications that are working.

That last one is not a pharmacologic effect. It still changes outcomes.