Why GLP-1 Trials Use Different Endpoints

When you read that a GLP-1 drug “worked” in a trial, the first question worth asking is: worked at what? A study designed to measure blood-sugar control answers a different question than one designed to measure weight loss, and both differ from one built to count heart attacks and strokes. The endpoint a trial chooses is not a technicality. It defines what the result can honestly claim.

The three families of endpoints

Most GLP-1 and dual-agonist trials are built around one of three primary measures.

Glycemic endpoints: usually change in HbA1c, the marker of average blood sugar over months. These were the original endpoints for diabetes drugs and remain the regulatory backbone for diabetes approval.
Weight endpoints: percent change in body weight. SURMOUNT-1, for example, reported up to 22.5% average weight loss with tirzepatide over 72 weeks. A drug can post impressive weight numbers without that automatically translating to other health gains.
Cardiovascular outcome endpoints: counts of “hard” events like heart attack, stroke, or cardiovascular death, usually grouped into a composite. These trials are larger, longer, and far more expensive, because events take years to accumulate.

A surrogate endpoint like A1c or weight tells you the drug moves a number. An outcomes endpoint tells you the drug changes what happens to people. They are not interchangeable.

Why the distinction matters

Vegetables, fruits, food — illustrating Why GLP-1 Trials Use Different Endpoints

A surrogate endpoint is a stand-in for what we care about. Lower A1c is meaningful because, over time, it usually tracks with fewer complications. But “usually” is doing real work in that sentence. Drug history is full of agents that improved a surrogate marker while failing, or even harming, on outcomes. That is why cardiovascular outcome trials became the gold standard: they test whether the surrogate improvement buys a longer, healthier life.

The SELECT trial is the cleanest recent example. It enrolled 17,604 adults with overweight or obesity and established cardiovascular disease but no diabetes, and asked an outcomes question directly. Semaglutide 2.4 mg reduced the composite of cardiovascular death, heart attack, or stroke by 20% versus placebo (6.5% vs 8.0% of patients; hazard ratio 0.80), published in the New England Journal of Medicine in 2023. That is a result about events, not just a number on a scale.

How to read a headline responsibly

When a result crosses your feed, it helps to mentally tag it: Watermelon, melon, colorful — illustrating Why GLP-1 Trials Use Different Endpoints

Was the primary endpoint a surrogate (weight, A1c) or an outcome (events, mortality)?
Was the comparison against placebo or against another active drug?
Was the population studied similar to the person being prescribed the drug?

The newer GLP-1 and dual agonists increasingly have outcome data behind some uses, which is genuinely notable. But not every approved use rests on outcomes evidence, and marketing rarely makes the distinction clear.

The takeaway

The endpoint is the question. A weight-loss number and a reduced-stroke number are both valuable, but they answer different things, and conflating them overstates what we know. Read any GLP-1 claim by first identifying what was measured, and treat surrogate wins as promising rather than proven until outcomes confirm them.