GLP-1 research roundup: what the 2025 trials actually showed

TLDR. The 2025 GLP-1 evidence base is wider than weight loss. Semaglutide has randomized-trial wins in cardiovascular disease (SELECT, 20% fewer major events), kidney disease (FLOW, 24% fewer kidney events) and biopsy-confirmed liver disease (ESSENCE, MASH resolution in 62.9% vs 34.3%). A large observational cohort tied GLP-1 use to roughly 17% lower overall cancer rates. The cautionary side is just as real: a well-powered Alzheimer's trial (evoke) found no benefit, an observational eye study flagged a more-than-tripled risk of a rare optic-nerve stroke, and the cancer and eye findings are not randomized. Below, each result with its design, its size, and the line it does not cross.

Sources are cited inline in each section. Randomized controlled trials carry more weight than observational cohorts, which can show association but cannot establish that the drug caused the outcome.

Cardiovascular: the SELECT trial

SELECT randomized 17,604 adults who had obesity and established cardiovascular disease but not diabetes to semaglutide 2.4mg or placebo. Major adverse cardiovascular events fell 20% on the drug (NEJM, Lincoff, November 11 2023). This is the result that earned Wegovy its FDA cardiovascular indication in March 2024.

The limit is in the enrollment criteria. Every participant already had heart disease, so the 20% reduction is a secondary-prevention finding. It does not show that semaglutide prevents a first cardiac event in someone with no prior diagnosis. For how that indication changes coverage, see our SELECT trial deep-dive.

Kidney: the FLOW trial

FLOW randomized 3,533 patients who had type 2 diabetes and chronic kidney disease. Major kidney events fell 24% on semaglutide, and the trial was stopped early for efficacy (NEJM, Perkovic, May 24 2024).

Two facts cap how far the result travels. Every participant had diabetes, and every participant was already on a renin-angiotensin inhibitor, the standard kidney-protective drug class. So FLOW shows benefit on top of standard care in diabetic kidney disease. It does not extend to non-diabetic chronic kidney disease, which was not studied. Our FLOW trial explainer covers the renal cautions in full.

Liver: the ESSENCE trial

ESSENCE tested semaglutide against placebo in patients with biopsy-confirmed MASH, the inflammatory form of fatty liver disease formerly called NASH. At 72 weeks, steatohepatitis resolved in 62.9% on semaglutide versus 34.3% on placebo, read on liver biopsy (NEJM, June 5 2025).

The caveats are about endpoint type and stage. This was an interim look at part of a larger 1,197-patient trial, and resolution-on-biopsy is a histology surrogate. It is a reasonable proxy for liver health, but it is not a hard outcome such as cirrhosis, liver transplant or death. Those outcomes have not yet been reported. For the related tirzepatide data, see Zepbound for MASH and fatty liver.

Cancer: the JAMA Oncology cohort

An observational study of 43,317 GLP-1 users linked the drugs to roughly 17% lower overall cancer rates, a hazard ratio of 0.83, with the strongest associations for endometrial, ovarian and meningioma cancers (JAMA Oncology, August 21 2025).

This one needs the heaviest caution of the positive findings. It is observational, not randomized, so it can show association but cannot prove the drug lowered cancer risk. People who take GLP-1 drugs and stay on them may differ from those who do not in ways that also affect cancer. The same study flagged a nonsignificant increase in kidney cancer, which is a reason to follow the data rather than treat the headline number as settled.

The Alzheimer's miss: evoke and evoke+

Not every hypothesis held up. The evoke and evoke+ trials randomized 3,808 adults with early Alzheimer's to oral semaglutide or placebo. The drug did not slow disease progression versus placebo. Novo Nordisk reported the result and discontinued the open-label extension (Novo Nordisk / Alzheimer Europe, November 24 2025).

This was a well-powered, clean negative, and it is worth stating plainly because the earlier observational signals had been promising. A randomized trial designed to answer the question said no. That carries more weight than the cohort hints that preceded it.

The vision question: NAION

A single-center cohort of 16,827 patients tied semaglutide to a more-than-tripled risk of non-arteritic anterior ischemic optic neuropathy (NAION), a rare stroke of the optic nerve. Among diabetic patients, the cumulative incidence was 8.9% on semaglutide versus 1.8% off it (JAMA Ophthalmology, Hathaway and Rizzo, July 3 2024).

The reasons for caution run in both directions. The study was observational and drawn from a single neuro-ophthalmology referral center, which introduces referral bias toward patients with eye problems. Larger follow-up studies have been inconsistent, and the absolute risk of NAION remains low even at the elevated relative rate. The signal is real enough to track and is driving litigation, but it is not established cause and effect.

How to read this body of evidence

The randomized trials, SELECT, FLOW, ESSENCE and evoke, are the load-bearing results, and they point in different directions: clear cardiovascular and kidney benefit in specific populations, encouraging but not-yet-confirmed liver benefit, and no Alzheimer's benefit. The observational findings, the cancer cohort and the NAION cohort, generate hypotheses worth following but cannot stand on their own.

For the weight-loss efficacy numbers that anchor most prescribing decisions, see our registration-trials explainer. For the day-to-day tolerability picture, see GLP-1 side effects, ranked by frequency.

Frequently asked questions

Do GLP-1 drugs reduce cancer risk?

One observational study of 43,317 users found roughly 17% lower overall cancer rates, strongest for endometrial, ovarian and meningioma cancers (JAMA Oncology, August 21 2025). Because it is observational and not randomized, it shows an association rather than proof of cause. The same study also flagged a nonsignificant increase in kidney cancer, so the finding is a reason to follow the research, not a treatment claim.

Does semaglutide protect the heart in people without existing heart disease?

The SELECT trial showed a 20% reduction in major cardiovascular events, but only in adults who already had established cardiovascular disease (NEJM, November 11 2023). There is no randomized evidence that it prevents a first cardiac event in someone with no prior diagnosis. The result is secondary prevention, not primary prevention.

Is the GLP-1 vision risk confirmed?

No. A single-center cohort of 16,827 patients reported a more-than-tripled risk of NAION, a rare optic-nerve stroke (JAMA Ophthalmology, July 3 2024), but the study was observational, subject to referral bias, and larger follow-up studies have been inconsistent. The absolute risk is low. The signal is being tracked and has prompted litigation, but cause and effect is not established.

Did the Alzheimer's trial work?

No. The evoke and evoke+ trials randomized 3,808 adults with early Alzheimer's and found that oral semaglutide did not slow progression versus placebo. Novo Nordisk discontinued the extension (November 24 2025). It was a well-powered negative result.

Does the kidney benefit apply to people without diabetes?

The FLOW trial showed a 24% reduction in major kidney events (NEJM, May 24 2024), but every participant had type 2 diabetes and was already taking a renin-angiotensin-system inhibitor. The result does not extend to non-diabetic chronic kidney disease, which the trial did not study.