Semaglutide in Human Clinical Trials: Populations, Endpoints and Findings

A trial-by-trial account of the published SUSTAIN, PIONEER, STEP, SELECT, FLOW, SOUL, ESSENCE and evoke programmes: populations, designs, durations, primary endpoints, results as reported, adverse events and limitations.

By Published Updated

Semaglutide is an acylated analogue of glucagon-like peptide-1, designed for once-weekly subcutaneous administration and later reformulated for once-daily oral administration [1]. Its clinical record is unusual in this field for two reasons: it extends across more than a decade, and a substantial part of it consists of event-driven outcome trials that count deaths, myocardial infarctions, strokes and kidney failure rather than surrogate measures.

This article describes that record programme by programme. For each trial it states the population enrolled, the design, the duration, the primary endpoint as the protocol defined it, the results as the publication reports them, the adverse events described, and the limitations that remain. It includes a phase 3 programme that did not meet its endpoint, because an evidence base is defined as much by its negative trials as by its positive ones.

Two boundaries apply throughout. Every study below tested a pharmaceutical product — material manufactured to a regulatory standard, administered under clinical supervision, in screened populations, under protocols registered before enrolment. And nothing below is guidance of any kind on handling any material, a comparison of products, or a recommendation.

The molecule these trials tested

In vitro research

Semaglutide was described in 2015 as a GLP-1 analogue engineered for extended duration: substitution at position 8 to resist dipeptidyl peptidase-4 cleavage, substitution at position 34, and acylation with a C18 fatty diacid through a short linker at position 26 [1]. The acylation is what binds the peptide reversibly to serum albumin, protecting it from proteolysis and renal filtration and producing a half-life of approximately one week.

That single design decision is the reason the trials below could be run at the durations they were. A peptide cleared in minutes cannot be studied over 104 weeks against a cardiovascular endpoint.

SUSTAIN-6 and PIONEER-6: the cardiovascular safety trials

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

SUSTAIN-6. 3,297 patients with type 2 diabetes on standard care were randomised to once-weekly semaglutide 0.5 mg or 1.0 mg or placebo for 104 weeks. At baseline 83.0% had established cardiovascular disease, chronic kidney disease or both. The primary composite outcome was first occurrence of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke, tested for noninferiority against a margin of 1.8 [2, 11].

The primary outcome occurred in 108 of 1,648 (6.6%) on semaglutide and 146 of 1,649 (8.9%) on placebo — hazard ratio 0.74, 95% CI 0.58 to 0.95, P<0.001 for noninferiority. Nonfatal myocardial infarction occurred in 2.9% against 3.9% (P=0.12) and nonfatal stroke in 1.6% against 2.7% (P=0.04); cardiovascular death rates were similar. New or worsening nephropathy was less frequent on semaglutide. Retinopathy complications were significantly more frequent on semaglutide — vitreous haemorrhage, blindness, or conditions requiring intravitreal treatment or photocoagulation — at a hazard ratio of 1.76 (95% CI 1.11 to 2.78; P=0.02) [2].

PIONEER-6. 3,183 patients at high cardiovascular risk were randomised to once-daily oral semaglutide or placebo in an event-driven noninferiority trial with the same composite primary outcome and the same 1.8 margin. Median time in trial was 15.9 months. Events occurred in 61 of 1,591 (3.8%) against 76 of 1,592 (4.8%), a hazard ratio of 0.79 (95% CI 0.57 to 1.11; P<0.001 for noninferiority) [3, 12].

Limitations. Both trials were designed to exclude excess risk, not to demonstrate benefit, and both were powered accordingly. The SUSTAIN-6 result is frequently quoted as a demonstration of cardiovascular benefit; the trial's own stated hypothesis was noninferiority. PIONEER-6 was short for an outcome trial at a median 15.9 months, and its confidence interval crosses 1. The retinopathy finding in SUSTAIN-6 is a genuine safety signal in the published record and is reported less often than the composite.

The STEP programme in overweight and obesity

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

STEP 1. 1,961 adults with a body-mass index of 30 or greater, or 27 or greater with at least one weight-related coexisting condition, and without diabetes, were randomised 2:1 to 68 weeks of once-weekly subcutaneous semaglutide 2.4 mg or placebo, both with lifestyle intervention. Coprimary endpoints were percentage change in body weight and a reduction of at least 5% [4, 13].

Mean change in body weight to week 68 was −14.9% against −2.4%, an estimated treatment difference of −12.4 percentage points (95% CI −13.4 to −11.5; P<0.001), or −15.3 kg against −2.6 kg. Reductions of 5%, 10% and 15% or more occurred in 86.4%, 69.1% and 50.5% of semaglutide participants against 31.5%, 12.0% and 4.9% of placebo participants. Nausea and diarrhoea were the most common adverse events, typically transient and mild to moderate; 4.5% of semaglutide participants discontinued for gastrointestinal events against 0.8% of placebo participants [4].

STEP 2. 1,210 adults with a body-mass index of at least 27 kg/m², glycated haemoglobin of 7–10% and type 2 diabetes diagnosed at least 180 days before screening were recruited from 149 clinics in 12 countries and randomised 1:1:1 to semaglutide 2.4 mg, semaglutide 1.0 mg or placebo for 68 weeks, in a double-blind, double-dummy superiority trial [5, 14].

Estimated change in mean body weight to week 68 was −9.6% with 2.4 mg against −3.4% with placebo, a treatment difference of −6.2 percentage points (95% CI −7.3 to −5.2; p<0.0001). Reductions of at least 5% occurred in 68.8% against 28.5% (odds ratio 4.88). Adverse events were reported in 87.6% of the 2.4 mg group, 81.8% of the 1.0 mg group and 76.9% of the placebo group; gastrointestinal events, mostly mild to moderate, in 63.5%, 57.5% and 34.3% respectively [5].

Limitations. Both endpoints are anthropometric; neither trial counts a clinical event. Both ran 68 weeks and describe nothing beyond that, including what follows discontinuation. STEP 2's effect size is materially smaller than STEP 1's, in a population with diabetes — a consistent pattern across this drug family that the trials document without explaining. Both trials were conducted alongside a structured lifestyle intervention delivered to every arm.

SELECT: cardiovascular outcomes in the absence of diabetes

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

Population, design and endpoint. 17,604 patients aged 45 or older with pre-existing cardiovascular disease and a body-mass index of 27 or greater, with no history of diabetes, randomised 1:1 to once-weekly subcutaneous semaglutide 2.4 mg or placebo in a multicentre, double-blind, event-driven superiority trial. The primary endpoint was a composite of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke. Mean exposure was 34.2 months and mean follow-up 39.8 months [6, 15].

Results as reported. A primary event occurred in 569 of 8,803 (6.5%) on semaglutide and 701 of 8,801 (8.0%) on placebo — hazard ratio 0.80, 95% CI 0.72 to 0.90, P<0.001. Adverse events leading to permanent discontinuation of trial product occurred in 1,461 patients (16.6%) on semaglutide and 718 (8.2%) on placebo, P<0.001 [6].

Limitations. The population is secondary prevention: everyone enrolled already had cardiovascular disease, and the result does not transfer to primary prevention. The discontinuation figure — one in six on semaglutide, twice the placebo rate — is part of the same result and belongs beside the hazard ratio. The trial does not separate how much of the benefit is attributable to changes in body weight and how much to other mechanisms, and it was not designed to.

FLOW: kidney outcomes in type 2 diabetes

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

Population, design and endpoint. 3,533 patients with type 2 diabetes and chronic kidney disease, defined by estimated glomerular filtration rate and urinary albumin-to-creatinine ratio thresholds, randomised to subcutaneous semaglutide 1.0 mg weekly or placebo. The primary outcome was major kidney disease events: a composite of kidney failure (dialysis, transplantation or an estimated glomerular filtration rate below 15), at least a 50% reduction in that rate from baseline, or death from kidney-related or cardiovascular causes. Median follow-up was 3.4 years [7, 16].

Results as reported. The trial was stopped early on the recommendation of a prespecified interim analysis. Primary-outcome risk was 24% lower on semaglutide — 331 against 410 first events, hazard ratio 0.76, 95% CI 0.66 to 0.88, P=0.0003. The kidney-specific composite gave a hazard ratio of 0.79 (0.66 to 0.94) and cardiovascular death 0.71 (0.56 to 0.89). All confirmatory secondary outcomes favoured semaglutide: the mean annual filtration-rate slope was less steep by 1.16 mL per minute per 1.73 m² (P<0.001), major cardiovascular events were 18% less frequent (hazard ratio 0.82; P=0.029) and death from any cause 20% less frequent (0.80; P=0.01). Serious adverse events were reported in 49.6% on semaglutide against 53.8% on placebo [7].

Limitations. Early cessation at an interim analysis tends to produce effect estimates larger than a completed trial would. The enrolled population sits within specific filtration-rate and albuminuria bands, so the result is a statement about that band of chronic kidney disease. The amount administered — 1.0 mg — is the diabetes amount rather than the 2.4 mg studied in the obesity trials, and the two are not interchangeable in reading results across programmes.

SOUL: oral administration against a cardiovascular endpoint

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

Population, design and endpoint. 9,650 participants aged 50 or older with type 2 diabetes, glycated haemoglobin of 6.5% to 10.0%, and known atherosclerotic cardiovascular disease, chronic kidney disease or both, randomised to once-daily oral semaglutide up to 14 mg or placebo in addition to standard care, in a double-blind, event-driven superiority trial. The primary outcome was the same three-part major adverse cardiovascular event composite. Mean follow-up was 47.5 months [8].

Results as reported. A primary event occurred in 579 of 4,825 (12.0%; 3.1 events per 100 person-years) on oral semaglutide and 668 of 4,825 (13.8%; 3.7 per 100 person-years) on placebo — hazard ratio 0.86, 95% CI 0.77 to 0.96, P=0.006. The confirmatory secondary outcomes, including the five-part major kidney disease composite, did not differ significantly between groups. Serious adverse events occurred in 47.9% against 50.3%, and gastrointestinal disorders in 5.0% against 4.4% [8].

Limitations. The primary result is positive and the secondary hierarchy is not, which bounds what the trial establishes to the composite itself. The kidney composite result here sits alongside FLOW's positive kidney result in a different population with a different formulation and a different amount, and the two should not be merged. Oral bioavailability of a peptide is low and variable, which makes exposure less uniform across participants than in the subcutaneous trials.

ESSENCE: biopsy endpoints in steatohepatitis

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

Population, design and endpoint. 1,197 patients with biopsy-defined metabolic dysfunction-associated steatohepatitis and fibrosis stage 2 or 3, randomised 2:1 to once-weekly subcutaneous semaglutide 2.4 mg or placebo for 240 weeks. A planned interim analysis at week 72 in the first 800 patients is what has been published. The part-1 primary endpoints were resolution of steatohepatitis without worsening of fibrosis, and reduction in fibrosis without worsening of steatohepatitis [9, 17].

Results as reported. Resolution without worsening of fibrosis occurred in 62.9% of 534 semaglutide patients and 34.3% of 266 placebo patients — estimated difference 28.7 percentage points, 95% CI 21.1 to 36.2, P<0.001. Reduction in fibrosis without worsening occurred in 36.8% against 22.4% (difference 14.4 points; 7.5 to 21.3; P<0.001). Combined resolution and fibrosis reduction occurred in 32.7% against 16.1%. Mean change in body weight was −10.5% against −2.0%. Mean changes in bodily pain scores did not differ significantly. Gastrointestinal adverse events were more common on semaglutide [9].

Limitations. This is an interim analysis of a 240-week trial reported at 72 weeks in two thirds of the enrolled population; the trial has not completed. Histological endpoints are read by pathologists and carry sampling and reader variability. Crucially, a histological endpoint is still a surrogate: whether these changes translate into fewer hepatic decompensations, transplants or deaths is what the remaining 168 weeks are intended to address, and that has not been reported.

evoke and evoke+: a phase 3 programme that did not meet its endpoint

Human clinical research

Results from pharmaceutical clinical trials describe the investigational material and populations used in those studies and should not be interpreted as establishing the effects of research-grade materials offered for laboratory use.

Population, design and endpoint. Two multicentre, randomised, double-blind, placebo-controlled phase 3 trials across 566 sites in 40 countries, in participants aged 55 to 85 with amyloid-confirmed Alzheimer's disease and either mild cognitive impairment or mild dementia; evoke+ additionally included participants with significant small-vessel pathology. Assignment was 1:1 to once-daily oral semaglutide 14 mg or placebo for up to 156 weeks. The primary endpoint was change in the Clinical Dementia Rating–Sum of Boxes score from baseline to week 104 [10].

Results as reported. 9,981 participants were screened and 3,808 randomised between 18 May 2021 and 8 September 2023 — 1,855 in evoke and 1,953 in evoke+. Mean age was 72.2 years and mean baseline score 3.7. Mean change in the score to week 104 was 2.3 and 2.2 on semaglutide against 2.3 and 2.1 on placebo, giving estimated differences of −0.08 (95% CI −0.35 to 0.20; p=0.57) in evoke and 0.10 (−0.17 to 0.38; p=0.46) in evoke+. Treatment-emergent adverse events were reported in 91.2% on semaglutide against 84.8% on placebo. Five fatalities were considered treatment-related by investigators, one on semaglutide and four on placebo. Both trials were discontinued [10].

What this trial settles. Observational and mechanistic work had suggested a lower incidence of dementia after exposure to GLP-1 receptor agonists; two adequately powered randomised trials in 3,808 participants found no difference in clinical progression. This is the strongest available evidence on that question, and it is negative. It is also a useful calibration for the rest of this literature: a mechanism that looks plausible, and an association that looks consistent in observational data, can fail outright when randomised.

What the evidence base does not establish

That every trial in a programme succeeds. evoke and evoke+ did not [10], and the SOUL secondary hierarchy did not [8]. A drug with several positive outcome trials also has negative ones, and reading only the positive half misrepresents the record.

Head-to-head standing against the dual and triple agonists. Semaglutide has been the comparator in trials run by other sponsors rather than the subject of its own head-to-head programme against those compounds, and no trial has compared it with a triple receptor agonist.

Effects beyond the enrolled populations. Each trial recruited against narrow criteria — established cardiovascular disease, specific filtration-rate and albuminuria bands, biopsy-confirmed fibrosis stage 2 or 3, amyloid-confirmed early Alzheimer's disease. A result in one of those populations is a result in that population, and the trials provide no basis for extending it to people who would not have been enrolled.

What happens after discontinuation. No trial above followed participants through and beyond cessation as a primary question.

Why the effects occur. None of these trials was designed to separate the contribution of changes in body weight from direct receptor-mediated effects on the vasculature, the kidney or the liver. The mechanism of the outcome results is an open question in the published literature, not a settled one.

Safety signals in the record. The retinopathy finding in SUSTAIN-6 [2] and the 16.6% discontinuation rate in SELECT [6] are part of the published evidence base and are not superseded by later positive results.

All of the research described in this article is research into a pharmaceutical product, conducted by its sponsor under registered protocols, using material manufactured to a regulatory standard and administered under clinical supervision in defined populations. None of it is research into, or evidence about, research-grade material supplied for laboratory use.

Frequently Asked Questions

How many people have been enrolled in published semaglutide clinical trials?
The trials summarised here account for more than 40,000 randomised participants: 17,604 in SELECT [6], 9,650 in SOUL [8], 3,808 across evoke and evoke+ [10], 3,533 in FLOW [7], 3,297 in SUSTAIN-6 [2], 3,183 in PIONEER-6 [3], 1,961 in STEP 1 [4], 1,210 in STEP 2 [5] and 1,197 in ESSENCE [9]. The full development programme is larger.
Has semaglutide reduced cardiovascular events in a randomised trial?
In three. SUSTAIN-6 reported a composite of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke in 6.6% against 8.9% on placebo (hazard ratio 0.74) in type 2 diabetes [2]. SELECT reported 6.5% against 8.0% (hazard ratio 0.80) in cardiovascular disease without diabetes [6]. SOUL reported 12.0% against 13.8% (hazard ratio 0.86) with once-daily oral administration [8].
What did the STEP 1 trial report?
In 1,961 adults with a body-mass index of 30 or greater, or 27 or greater with a related condition and without diabetes, mean change in body weight over 68 weeks was −14.9% against −2.4% on placebo, an estimated treatment difference of −12.4 percentage points (95% CI −13.4 to −11.5; P<0.001). Reductions of 5% or more occurred in 86.4% against 31.5% [4].
Does semaglutide have published kidney outcome data?
Yes. FLOW randomised 3,533 participants with type 2 diabetes and chronic kidney disease and was stopped early at a prespecified interim analysis. The primary composite of kidney failure, a 50% reduction in estimated glomerular filtration rate, or kidney-related or cardiovascular death occurred 24% less often on semaglutide (hazard ratio 0.76; 95% CI 0.66 to 0.88; P=0.0003) over a median 3.4 years [7].
What happened in the semaglutide Alzheimer's disease trials?
They did not meet their primary endpoint. evoke and evoke+ randomised 3,808 participants with amyloid-confirmed early Alzheimer's disease to once-daily oral semaglutide 14 mg or placebo. Change in the Clinical Dementia Rating-Sum of Boxes score at week 104 did not differ from placebo (estimated difference −0.08, p=0.57 in evoke; 0.10, p=0.46 in evoke+), and both trials were discontinued [10].
What adverse events are reported in semaglutide trials?
Gastrointestinal events, chiefly nausea and diarrhoea, are the most frequently reported and are described as typically transient and mild to moderate [4]. Discontinuation attributed to adverse events was 16.6% against 8.2% on placebo in SELECT [6]. SUSTAIN-6 reported a significantly higher rate of retinopathy complications on semaglutide (hazard ratio 1.76; 95% CI 1.11 to 2.78; P=0.02) [2].
How long have semaglutide clinical trials run?
SUSTAIN-6 ran 104 weeks [2], the STEP trials 68 weeks [4, 5], and the event-driven outcome trials longer: mean follow-up was 39.8 months in SELECT [6], 47.5 months in SOUL [8] and a median 3.4 years in FLOW [7]. ESSENCE is a 240-week trial reported at a 72-week interim analysis [9], and evoke followed participants for up to 156 weeks [10].

References

  1. Discovery of the Once-Weekly Glucagon-Like Peptide-1 (GLP-1) Analogue Semaglutide Journal of Medicinal Chemistry; 2015. PMID 26308095 doi:10.1021/acs.jmedchem.5b00726
  2. Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes The New England Journal of Medicine; 2016. PMID 27633186 doi:10.1056/NEJMoa1607141
  3. Oral Semaglutide and Cardiovascular Outcomes in Patients with Type 2 Diabetes The New England Journal of Medicine; 2019. PMID 31185157 doi:10.1056/NEJMoa1901118
  4. Once-Weekly Semaglutide in Adults with Overweight or Obesity The New England Journal of Medicine; 2021. PMID 33567185 doi:10.1056/NEJMoa2032183
  5. Semaglutide 2·4 mg once a week in adults with overweight or obesity, and type 2 diabetes (STEP 2): a randomised, double-blind, double-dummy, placebo-controlled, phase 3 trial Lancet; 2021. PMID 33667417 doi:10.1016/S0140-6736(21)00213-0
  6. Semaglutide and Cardiovascular Outcomes in Obesity without Diabetes The New England Journal of Medicine; 2023. PMID 37952131 doi:10.1056/NEJMoa2307563
  7. Effects of Semaglutide on Chronic Kidney Disease in Patients with Type 2 Diabetes The New England Journal of Medicine; 2024. PMID 38785209 doi:10.1056/NEJMoa2403347
  8. Oral Semaglutide and Cardiovascular Outcomes in High-Risk Type 2 Diabetes The New England Journal of Medicine; 2025. PMID 40162642 doi:10.1056/NEJMoa2501006
  9. Phase 3 Trial of Semaglutide in Metabolic Dysfunction-Associated Steatohepatitis The New England Journal of Medicine; 2025. PMID 40305708 doi:10.1056/NEJMoa2413258
  10. Efficacy and safety of oral semaglutide 14 mg (flexible dose) in early-stage symptomatic Alzheimer's disease (evoke and evoke+): two phase 3, randomised, placebo-controlled trials Lancet; 2026. PMID 41865758 doi:10.1016/S0140-6736(26)00459-9
  11. Trial to Evaluate Cardiovascular and Other Long-term Outcomes With Semaglutide in Subjects With Type 2 Diabetes. NCT01720446
  12. A Trial Investigating the Cardiovascular Safety of Oral Semaglutide in Subjects With Type 2 Diabetes. NCT02692716
  13. STEP 1: Research Study Investigating How Well Semaglutide Works in People Suffering From Overweight or Obesity. NCT03548935
  14. Research Study Investigating How Well Semaglutide Works in People With Type 2 Diabetes Suffering From Overweight or Obesity. NCT03552757
  15. Semaglutide Effects on Heart Disease and Stroke in Patients With Overweight or Obesity. NCT03574597
  16. A Research Study to See How Semaglutide Works Compared to Placebo in People With Type 2 Diabetes and Chronic Kidney Disease. NCT03819153
  17. Research Study on Whether Semaglutide Works in People With Non-alcoholic Steatohepatitis (NASH). NCT04822181

Every identifier above is resolved against PubMed, Crossref or ClinicalTrials.gov at build time, and the title returned by the register is compared with the title stored here. A page does not publish if a reference fails to resolve.

Research-Use Information

For in vitro research use only. This material is a laboratory reagent. It is not a drug, food, dietary supplement, or cosmetic and is not for human or veterinary use, including ingestion, injection, or any other administration. No information on this page describes or implies any effect in humans or animals. Sold only to researchers under our Terms of Sale.