Clinical trials

What the Clinical Trials Actually Show About Semaglutide

A plain-English tour of the roughly 40 major semaglutide trials — what each one measured, what the numbers were, and where the evidence runs out.

Last verified September 14, 2026·19 sources cited

Semaglutides.org is for information only and is not medical advice. Always talk to a licensed healthcare provider about your own care. Some links to telehealth services are affiliate links, labeled where they appear.

Semaglutide is one of the most heavily studied medicines of the last decade. That is a good thing for anyone trying to make a decision about it, because there is real data to look at instead of marketing claims. The problem is that the data is spread across more than 40 phase 3 trials with names like SUSTAIN, PIONEER, STEP, OASIS, SELECT, FLOW and SOUL, and almost nobody outside of medicine can keep them straight.

This page walks through what each group of trials tested, what the numbers were, and — just as important — where the evidence stops. Nothing here is medical advice. Every treatment decision belongs in a conversation with a healthcare provider who knows your history.

What are the four main semaglutide trial programs?

Semaglutide is the active ingredient in four different products, and each one has its own body of evidence.

SUSTAIN tested the weekly injection in type 2 diabetes. This is the program behind Ozempic. It ran from SUSTAIN 1 through SUSTAIN 11, plus SUSTAIN FORTE, and compared semaglutide against placebo, sitagliptin, exenatide, insulin glargine, dulaglutide, canagliflozin, liraglutide and mealtime insulin.

PIONEER tested the daily tablet in type 2 diabetes. This is the program behind Rybelsus, covering PIONEER 1 through 10 plus PIONEER PLUS.

STEP tested the higher 2.4 mg weekly injection for weight management. This is the program behind Wegovy, running from STEP 1 through STEP 10, plus STEP TEENS, two heart failure trials, and the newer STEP UP trials of a 7.2 mg dose.

OASIS tested high-dose tablets for weight management. OASIS 4 supported the FDA approval of the Wegovy tablet in December 2025.

Layered on top are the outcome trials — SELECT for heart events, FLOW for kidney disease, SOUL for heart events with the tablet, ESSENCE for liver disease, STRIDE for leg circulation, and evoke for Alzheimer’s disease.

What did the diabetes trials find?

The SUSTAIN program answered a straightforward question: does this drug lower blood sugar, and how does it compare with what people were already taking?

SUSTAIN 1 was the starting point. In 388 adults with newly diagnosed type 2 diabetes, HbA1c fell 1.45 to 1.55 percentage points over 30 weeks compared with essentially no change on placebo, and body weight fell 3.7 to 4.5 kg [1]. The head-to-head trials that followed were more interesting. Semaglutide beat sitagliptin, exenatide extended-release, insulin glargine, dulaglutide, canagliflozin and liraglutide on blood sugar, and it beat all of them on weight. Against insulin glargine, for example, participants lost 3.5 to 5.2 kg while the insulin group gained 1.2 kg.

The trial that mattered most was SUSTAIN 6. It enrolled 3,297 people with type 2 diabetes, most of whom already had heart or kidney disease, and tracked serious cardiovascular events for two years. Major events occurred in 6.6% of the semaglutide group versus 8.9% on placebo, a hazard ratio of 0.74 [2]. That was the first hint that this class of drug does more than lower a lab value.

SUSTAIN 6 also produced the clearest safety signal in the whole program: diabetic retinopathy complications were more common with semaglutide, 3.0% versus 1.8%. That finding is why the US label carries a warning about diabetic retinopathy, and why the mechanism is still being studied.

The PIONEER program did the same work for the tablet. The short version is that the 7 mg and 14 mg doses beat sitagliptin and empagliflozin on blood sugar, and the 14 mg dose matched a daily liraglutide injection while producing more weight loss. PIONEER 6 tested cardiovascular safety in 3,183 people and found a hazard ratio of 0.79, with fewer deaths from any cause, but the trial was short and not designed to prove benefit [3].

That gap sat open for six years until SOUL. Published in March 2025, SOUL followed 9,650 people with type 2 diabetes and either atherosclerotic disease, chronic kidney disease or both, for a median of about four years. Major adverse cardiovascular events occurred in 12.0% of the tablet group versus 13.8% on placebo — a hazard ratio of 0.86 [4]. That result made oral semaglutide the first GLP-1 pill proven to prevent heart attacks and strokes, and the FDA added the indication in October 2025.

What did the obesity trials find?

STEP 1 is the trial most people have heard about even if they do not know the name. It randomized 1,961 adults with obesity and without diabetes to semaglutide 2.4 mg or placebo for 68 weeks, alongside lifestyle support. Average weight change was -14.9% versus -2.4% on placebo [5]. In absolute terms that was 15.3 kg, or about 34 pounds, from an average starting weight of 105 kg.

The distribution matters more than the average. In STEP 1, 86.4% of participants lost at least 5% of their body weight, 69.1% lost at least 10%, and 50.5% lost at least 15%. That also means roughly one in seven people did not reach even 5%. Response varies a great deal between individuals, and no trial can predict which group any particular person falls into.

Several STEP trials tested the same drug in different situations:

STEP 2 enrolled people with type 2 diabetes and found 9.6% weight loss versus 3.4% on placebo [6]. Weight loss is consistently smaller in people with diabetes, which holds true for tirzepatide as well.
STEP 3 added intensive behavioral therapy — 30 counseling visits and an initial low-calorie diet — and reached 16.0% [7].
STEP 5 ran two full years and reached 15.2%, showing the weight loss holds rather than reversing while treatment continues [8].
STEP 8 compared semaglutide with liraglutide 3.0 mg head-to-head: 15.8% versus 6.4%, with half the dropout rate.
STEP TEENS tested adolescents aged 12 to 17 and found BMI fell 16.1% versus a 0.6% increase on placebo.
STEP UP, published in September 2025, tested a higher 7.2 mg weekly dose in 1,407 adults and reported 18.7% weight loss versus 15.6% for the standard 2.4 mg dose [9]. The FDA approved that higher dose as Wegovy HD in March 2026.

What do the outcome trials show beyond weight and blood sugar?

This is where semaglutide’s evidence base separates itself from older weight-loss drugs.

SELECT is the big one. It enrolled 17,604 adults aged 45 and older who had established cardiovascular disease and a BMI of 27 or higher, but no diabetes. Over an average follow-up of about 40 months, the combined rate of cardiovascular death, nonfatal heart attack and nonfatal stroke was 6.5% on semaglutide versus 8.0% on placebo — a hazard ratio of 0.80 [10]. That translated into the FDA cardiovascular risk-reduction indication for Wegovy in 2024, which in turn opened a path to Medicare Part D coverage for that use.

FLOW tested the 1.0 mg dose in 3,533 people with type 2 diabetes and chronic kidney disease. Major kidney events — kidney failure, a 50% drop in filtration rate, or death from kidney or cardiovascular causes — fell 24% (hazard ratio 0.76), and death from any cause fell 20% [11]. The trial was stopped early because the benefit was so clear. The FDA added a kidney indication to Ozempic in January 2025.

ESSENCE tested semaglutide in metabolic dysfunction-associated steatohepatitis, or MASH, the inflammatory form of fatty liver disease. In the first 800 patients analyzed at 72 weeks, liver inflammation resolved without worsening scarring in 62.9% of the semaglutide group versus 34.3% on placebo, and scarring improved in 36.8% versus 22.4% [12]. That earned an accelerated approval in August 2025. Accelerated approval means the drug was cleared on a surrogate measure — liver biopsy findings — with a confirmatory trial still running. The second part of ESSENCE, which will test whether people actually avoid cirrhosis, is expected to read out around 2029.

STRIDE tested whether semaglutide helps people with type 2 diabetes and painful leg circulation problems walk farther. Maximum treadmill walking distance improved about 13% more than placebo over 52 weeks [13].

STEP 9 looked at knee osteoarthritis. In 407 adults with obesity and moderate knee arthritis, pain scores fell 41.7 points versus 27.5 on placebo on a 100-point scale [14]. STEP 10 found that 81% of people with obesity and prediabetes returned to normal blood sugar after a year, versus 14% on placebo [15].

Where has semaglutide failed?

The most important failure is Alzheimer’s disease. Observational studies had suggested that people taking GLP-1 drugs developed dementia less often, and Novo Nordisk ran two enormous trials — evoke and evoke+ — testing once-daily oral semaglutide up to 14 mg (not the Wegovy injection) in 3,808 people with amyloid-confirmed early Alzheimer’s disease across 566 sites in 40 countries.

The results were flatly negative. At 104 weeks, the change in the standard dementia rating scale was 2.3 points on semaglutide and 2.3 points on placebo in evoke, and 2.2 versus 2.1 in evoke+. The estimated differences were -0.08 (95% CI -0.35 to 0.20, p=0.57) and 0.10 (95% CI -0.17 to 0.38, p=0.46) [16]. Neither came close to significance. Some Alzheimer’s-related biomarkers improved, but that did not translate into any clinical benefit. Novo Nordisk announced the topline result on November 24, 2025 and discontinued the program including its one-year extension phases. Full results were presented in March 2026 and published in The Lancet.

It is worth sitting with that result, because it illustrates something general: observational associations between a drug and a good outcome frequently do not survive a randomized trial.

What do the trials say about side effects and dropouts?

Gastrointestinal side effects are the defining tolerability issue across every program. In STEP 1, nausea and diarrhea were the most common adverse events, described as typically transient, mild to moderate, and fading over time; 4.5% of the semaglutide group stopped because of gastrointestinal events versus 0.8% on placebo [5].

But that 4.5% figure understates the real-world picture in two ways. First, discontinuation for any reason is much higher. In SELECT, 16.6% of the semaglutide group permanently stopped the trial drug because of adverse events, versus 8.2% on placebo, and overall discontinuation reached 26.7% over roughly three years [10]. Second, trial participants get free medication, regular visits and structured support. Ordinary patients do not.

A Cleveland Clinic study of 7,881 people who started injectable semaglutide or tirzepatide for obesity in routine care found average one-year weight loss of 8.7% — well below the trial figures. The reason was mostly behavioral and logistical rather than biological: 80.8% never reached a high maintenance dose, and people who stopped within three months lost only 3.6%. Those who stayed on treatment lost 11.9%, and those who stayed on at a high maintenance dose lost 13.7%, close to trial performance [17].

What about muscle loss?

This is the question the trials answer least well. A body composition sub-study within STEP 1 scanned 140 participants at baseline and week 68. Total fat mass fell 19.3% and deep abdominal fat fell 27.4%, while total lean body mass fell 9.7% [18]. Published analyses of that data set put lean mass at roughly 39 to 45% of total weight lost.

Two caveats matter. Lean body mass on a DEXA scan includes water, organs and connective tissue, not just skeletal muscle — so a 9.7% drop in lean mass is not a 9.7% drop in muscle. And because fat fell faster than lean tissue, lean mass as a share of total body weight actually rose by three percentage points. Whether the absolute loss has functional consequences, especially in older adults, is an open research question. It is a reasonable thing to raise with a healthcare provider, particularly if you are over 65 or already have limited strength.

How should you read a trial number?

One practical skill is worth learning: the same trial will often be quoted with two different weight-loss figures.

The treatment policy estimand counts every randomized participant, whether or not they stayed on the drug. It answers “what happened to people assigned this treatment?”

The trial product estimand estimates what would have happened if everyone had stayed on treatment and taken nothing else. It answers “what does this drug do when you actually take it?”

The second number is always larger. STEP UP is the clearest example: the Lancet paper reports 18.7% weight loss under the treatment policy estimand, while Novo Nordisk’s press materials report 20.7% under the trial product estimand [9]. Neither figure is wrong. OASIS 4 shows the same split: 13.6% in the New England Journal of Medicine paper, 16.6% in the company announcement. When you see two numbers that seem to contradict each other, this is usually why.

How does semaglutide compare with everything else?

A network meta-analysis published in Nature Medicine in October 2025 pooled 56 randomized trials covering 60,307 patients across orlistat, liraglutide, naltrexone/bupropion, phentermine/topiramate, semaglutide and tirzepatide. Only semaglutide and tirzepatide exceeded 10% average total body weight loss [19].

The analysis also mapped which drug has proven which secondary benefit. Semaglutide is the only one with demonstrated reductions in major cardiovascular events and in knee osteoarthritis pain. Tirzepatide is the only one with demonstrated remission of obstructive sleep apnea. Both showed restoration of normal blood sugar, type 2 diabetes remission and reduced heart failure hospitalization. That is the honest state of the evidence: these two drugs are in a tier of their own, and neither one dominates the other across every outcome.

What is still unknown?

Several things. Nobody has run a randomized trial of tapering off semaglutide, so there is no evidence-based way to come off it. There is no head-to-head randomized comparison of the Wegovy tablet against the Wegovy injection. The confirmatory liver outcome data from ESSENCE Part 2 is years away. And there is no randomized evidence at all beyond four years, which is the length of SELECT — a meaningful limitation for a medicine that most evidence suggests works only while you take it.

Sources

Questions people ask

How many clinical trials has semaglutide been through?

More than 40 phase 3 trials across four main programs — SUSTAIN for the weekly injection in type 2 diabetes, PIONEER for the daily tablet in type 2 diabetes, STEP for the 2.4 mg obesity dose, and OASIS for the obesity tablet — plus outcome trials such as SELECT, FLOW, SOUL, ESSENCE and STRIDE. Together they include well over 60,000 randomized participants.

What is the biggest semaglutide trial?

SELECT, with 17,604 participants followed for an average of about 40 months. It tested whether semaglutide 2.4 mg reduces heart attacks and strokes in people with cardiovascular disease and excess weight but no diabetes.

Did semaglutide fail any trials?

Yes. The evoke and evoke+ trials tested once-daily oral semaglutide in 3,808 people with early Alzheimer's disease and found no slowing of cognitive or functional decline compared with placebo. Novo Nordisk announced the negative topline result in November 2025 and discontinued the program; the full results were published in The Lancet in 2026.

Does semaglutide cause muscle loss in the trials?

A 140-person scan sub-study within STEP 1 found total lean body mass fell 9.7% while fat mass fell 19.3%. In percentage terms, roughly 39 to 45% of weight lost was lean tissue, though lean mass as a share of total body weight still improved. Lean mass is not the same as skeletal muscle, and this remains an active research question.

How long do the semaglutide trials run?

Most obesity trials run 68 to 72 weeks. STEP 5 ran two years, and SELECT followed people for up to four years — the longest randomized evidence available.

Why do press releases and journal papers report different weight-loss numbers for the same trial?

They usually use different statistical approaches. The treatment policy estimand counts everyone who was randomized, including people who stopped the drug. The trial product estimand estimates what would happen if everyone stayed on treatment. The second number is always larger.

Who paid for the semaglutide trials?

Novo Nordisk funded nearly all of the phase 3 program, including SUSTAIN, PIONEER, STEP, OASIS, SELECT, FLOW, SOUL, ESSENCE, STRIDE and evoke. A small number of investigator-initiated studies, such as the alcohol use disorder trial at the University of North Carolina, were funded independently.

This article summarizes FDA labeling, published research and company information current as of September 14, 2026. It is not medical advice and does not replace a conversation with your own healthcare provider. How we research and verify.