The Semaglutide Evidence File From STEP 1 to SURMOUNT-5

Four published figures describe how much weight semaglutide produced in trials. Every one of those figures is accurate. None of them mean the same thing. The 14.9% that circulates most widely comes from one trial, at one dose, over 68 weeks, under one statistical rule. Apply a different rule to that same trial and it reports 16.9%. Move to a different trial and the number shifts again. Most pages quoting these percentages drop the trial, the dose, the population, the duration and the statistical method that produced them. This page puts all of that back. It works through the major semaglutide studies one at a time, says what each actually measured, and names which figure belongs to which claim. Where sources disagree or the evidence runs out, that is stated plainly, including one widely repeated figure this review could not confirm. Every clinical result here describes an approved medicine, tested as a specific pharmaceutical product.

The short answer on semaglutide

Semaglutide is a GLP-1 receptor agonist, a synthetic peptide that activates the receptor for a hormone the gut releases after eating. It reaches patients under several brands carrying different labels. Ozempic and Rybelsus are approved for type 2 diabetes. Wegovy covers chronic weight management, cardiovascular risk reduction and liver disease. Wegovy HD is a 7.2 mg strength inside the Wegovy injection approval, not a separate medicine.

In STEP 1, adults with obesity and no diabetes lost 14.9% of body weight by week 68, against 2.4% on placebo. That figure is the treatment-policy estimand. STEP UP tested the 7.2 mg dose and reported 18.7% at 72 weeks, against 3.9% on placebo, on the same estimand.

A higher STEP UP figure of 20.7% circulates widely. It could not be verified in the published paper, so this page does not use it.

Last updated 15 October 2026. This review checked the primary publications for STEP 1 through STEP 8, STEP TEENS and STEP UP. It also checked SELECT, FLOW, SUSTAIN-6, PIONEER 6, SOUL, ESSENCE, OASIS 4, SURMOUNT-5 and REDEFINE 1. Regulatory checks covered the current Wegovy and Ozempic prescribing information, FDA approval letters, and the Federal Register record on 503B compounding. Prepared by the PrimeTime BioLabs editorial team. A named medical reviewer has not yet been assigned.

What semaglutide is and how its brands differ

Semaglutide activates the GLP-1 receptor, a class B G protein-coupled receptor. Activation raises intracellular cAMP and produces four well-characterised effects. The pancreas releases more insulin, but only when blood glucose is already high. Less glucagon signals the liver to release sugar. The stomach empties its contents more slowly. Brain regions governing appetite receive altered signalling.

What semaglutide is. Semaglutide is a synthetic GLP-1 receptor agonist of 31 amino acids, sharing 94% homology with human GLP-1. Chemical modifications extend its half-life to roughly one week. It is sold as Ozempic and Rybelsus for type 2 diabetes, and as Wegovy for weight management, cardiovascular risk and liver disease. Each brand carries its own approved label.

Three chemical modifications define this particular molecule. An aminoisobutyric acid substitution sits at position 8 under standard GLP-1 numbering, which is the second residue of the peptide itself. That change blocks cleavage by the enzyme DPP-4. A C18 fatty-diacid chain attaches at lysine 26 through a gamma-glutamic acid linker and two spacer units. Arginine replaces lysine at position 34, which controls where the fatty chain attaches. The formula is C₁₈₇H₂₉₁N₄₅O₅₉, molecular weight 4113.58 g/mol, CAS registry number 910463-68-2.

One caution about numbering matters a great deal here. Sources that write "position 2" are counting the peptide sequentially, and sources that write "position 8" are using native GLP-1 numbering. The FDA label uses position 8. Mixing both conventions in one description produces an inconsistent sentence.

Differences between the brands are not cosmetic. Ozempic now names two products, an injection approved in 2017 and a tablet approved in December 2024, both for type 2 diabetes. Rybelsus shares its approval with those Ozempic tablets. Wegovy covers an injection approved in 2021 and a tablet approved in December 2025, which carries the 25 mg oral strength studied in obesity. Approved medicines, investigational products and research-grade materials sold under the same compound name occupy separate categories. Evidence generated on one does not transfer to another.

Why semaglutide weight loss figures disagree

Two trials can test the same drug at the same dose and report different numbers for honest reasons. The most overlooked reason is the estimand, which is the statistical question a trial chose to answer.

A treatment-policy estimand asks what happened to everyone randomised, counting people who stopped the drug, reduced their dose or added another treatment. It answers a practical question about a whole group. A trial-product estimand asks what would have happened if everyone had taken the drug as planned, with no discontinuation and no rescue therapy. It isolates the drug's effect under ideal conditions. Neither estimand is more correct than the other. They answer different questions, and the trial-product figure is almost always the larger of the two.

STEP 1 shows the gap clearly. Its treatment-policy result was 14.9% against 2.4% on placebo, a difference of 12.4 percentage points. Its trial-product result from the same participants was 16.9% against 2.4%. The paper reports an estimated treatment difference of 14.4 points there, which is a model-adjusted figure and not a raw subtraction. Both results appear in the same paper. Pages quoting 16.9% as though it were the headline figure are not wrong about the number, only about which question it answers.

Why the figures disagree. Semaglutide weight-loss percentages differ because trials differ. STEP 1 reported 14.9% at 68 weeks on the treatment-policy estimand, in adults without diabetes. STEP UP reported 18.7% at 72 weeks on the same estimand, at a higher dose. Different doses, populations, durations and statistical rules produce different numbers from the same drug.

Dose explains part of the spread, since 2.4 mg and 7.2 mg are not the same exposure. Population differences explain a good deal more. Adults with type 2 diabetes consistently lose less weight, which is why STEP 2 reported 9.6% where STEP 1 reported 14.9%. What the placebo group received matters too. STEP 3's placebo arm lost 5.7%, because those participants also received thirty behavioural therapy sessions and an eight-week low-calorie diet.

That last point creates a trap worth naming. Subtracting placebo from drug produces a "placebo-subtracted effect" that looks comparable across trials. It is not. STEP 1's 12.4-point gap and STEP 3's 10.3-point gap describe different experiments with different control conditions.

One rule covers all of it. Results from separate trials were never randomised against each other, so comparing them produces an indirect comparison. Indirect comparisons are weaker evidence than a head-to-head trial, and should never be presented as one.

Which figure to quote

For a general statement about semaglutide 2.4 mg in adults without diabetes, quote STEP 1's 14.9% at 68 weeks, naming the treatment-policy estimand. For the highest dose studied, quote STEP UP's 18.7% at 72 weeks on the same estimand. For a direct comparison against tirzepatide, quote only SURMOUNT-5. For anything concerning diabetes, quote STEP 2 and never STEP 1.

What the major STEP trials actually showed

The STEP programme tested subcutaneous semaglutide 2.4 mg across populations that differ in ways the headline numbers hide.

STEP 1 randomised 1,961 adults with overweight or obesity and no diabetes, over 68 weeks. Mean weight change was 14.9% against 2.4% on placebo under the treatment-policy estimand. Responder rates tell the same story. Half the semaglutide group lost at least 15% of body weight, against 4.9% on placebo. The authors noted a preponderance of women and White participants, and a trial population possibly more committed to weight loss than most.

STEP 2 enrolled 1,210 adults who had type 2 diabetes, with BMI at or above 27 and HbA1c between 7% and 10%. Treatment-policy weight change was 9.6% against 3.4% on placebo. A third arm tested the 1.0 mg diabetes dose and produced 7.0%. People using insulin were excluded, and baseline cardiovascular risk was relatively low. The reduced effect in diabetes is consistent across this drug class and should be expected.

STEP 3 added intensive behavioural therapy for everyone. Among 611 adults, semaglutide produced 16.0% against 5.7% on placebo. Both arms received thirty counselling visits and an initial low-calorie diet, which is why the placebo result is unusually large. The authors note that this design cannot separate the drug's contribution from the therapy's. Reading 16.0% as a drug-alone effect misstates what the trial measured.

STEP 4 is the trial most often described incorrectly. All 902 participants first took open-label semaglutide for 20 weeks and lost 10.6% on average. Only the 803 who tolerated escalation to 2.4 mg were then randomised, to continue the drug or switch to placebo for 48 further weeks. Those who continued lost another 7.9%. Those switched to placebo regained 6.9%. The comparison runs from week 20, not from baseline, and lasts 48 weeks, not 68.

STEP 5 followed 304 adults for 104 weeks, the longest weight-management trial in the programme. Treatment-policy weight change was 15.2% against 2.6% on placebo. Semaglutide participants measured 15.6% at week 52 and 15.2% at week 104, showing little regain while treatment continued. The sample was small, so the estimate carries wider uncertainty than STEP 1's.

STEP 8 compared semaglutide against an active drug. Among 338 adults, weekly semaglutide 2.4 mg produced 15.8% against 6.4% for daily liraglutide 3.0 mg. Liraglutide was the comparator that the primary endpoint tested. The pooled placebo arm of 85 participants was a supporting analysis only. Quoting the placebo contrast as the headline finding inverts what the trial was built to measure. The trial ran open-label, and the two drugs handled poor tolerance differently. Participants who could not tolerate top-dose semaglutide stayed on a lower dose, while those on liraglutide stopped and restarted titration. The authors flagged that asymmetry as a limitation.

STEP TEENS studied 201 adolescents aged 12 to under 18. Its primary endpoint was percentage change in BMI, unlike every adult trial in the programme. BMI fell 16.1% against a 0.6% rise on placebo. The equivalent placebo-subtracted body weight figure was 17.4 percentage points. Quoting 16.1% as a weight change is the most common error made about this trial. All but one participant had obesity, so the paper states that no conclusion about adolescents with overweight is possible. Seven weeks after treatment ended, BMI had already begun moving back toward baseline.

What STEP UP adds to the evidence

STEP UP tested whether a higher dose produced more weight loss. It randomised 1,407 adults with BMI at or above 30 and no diabetes, over 72 weeks. The 5:1:1 split covered semaglutide 7.2 mg, semaglutide 2.4 mg and placebo.

Treatment-policy results were 18.7% for 7.2 mg, 15.6% for 2.4 mg and 3.9% for placebo. The difference against placebo was 14.8 percentage points, and against 2.4 mg it was 3.1 points. Both of those differences reached statistical significance. Waist circumference fell 11.7 cm more than placebo.

Tolerability moved together with the dose. Gastrointestinal adverse events affected 70.8% of the 7.2 mg group, 61.2% at 2.4 mg and 42.8% on placebo. Dysaesthesia, an altered or unpleasant skin sensation, appeared in 22.9% at 7.2 mg, 6.0% at 2.4 mg and 0.5% on placebo. Those percentages use the safety population, which is slightly smaller than the randomised total. Serious adverse events were 6.8%, 10.9% and 5.5% respectively, and did not rise with dose.

The 20.7% figure. Many pages report 20.7% as STEP UP's result for the 7.2 mg group, describing it as a trial-product estimand. That figure could not be confirmed. The published paper's abstract names only the treatment-policy estimand, and the full text and appendix sit behind a paywall. The European Medicines Agency's own assessment report states that no separate trial-product estimand was defined for the body weight endpoint. The 20.7% appears in a conference presentation by the investigators and in a company announcement. Neither of those is the same as a verified result in the primary publication. Evidence status: unverified claim. This page uses 18.7%.

Two further cautions apply to this trial. A companion trial called STEP UP T2D tested the same 7.2 mg dose in people who also had type 2 diabetes. That trial reported 13.2%, not 18.7%. The FDA approved the 7.2 mg dose on 19 March 2026, as a new dosing regimen within the existing Wegovy approval. That approval is confined to adult weight reduction. It does not extend the 7.2 mg dose to cardiovascular risk reduction, to liver disease or to patients under 18.

Every major semaglutide trial, adjudicated

Method note. Each trial below was checked against its primary publication, located through PubMed, with the DOI recorded. Regulatory statements were checked against FDA approval letters and the current prescribing information on DailyMed and accessdata.fda.gov, and against the Federal Register. Checks were completed on 15 October 2026. Where a paper reported more than one estimand, both are shown and labelled as the paper labels them. Where a figure sits only in a supplementary appendix this review could not retrieve, the cell reads "not retrieved" and no number is supplied. Where a number appears only in a conference presentation or a company release, it is marked unverified and excluded from the article body. The "which figure to quote" column reflects this review's adjudication, not the authors' own recommendation.

Trial designs

• Trial: STEP 1 — Population: Overweight or obesity, no diabetes — N: 1,961 — Duration: 68 weeks — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: % weight change
• Trial: STEP 2 — Population: Type 2 diabetes, BMI ≥27, HbA1c 7-10% — N: 1,210 — Duration: 68 weeks — Formulation and trial dose: SC 2.4 mg and 1.0 mg weekly — Endpoint discussed: % weight change
• Trial: STEP 3 — Population: No diabetes, plus intensive behavioural therapy — N: 611 — Duration: 68 weeks — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: % weight change
• Trial: STEP 4 — Population: No diabetes, tolerated 2.4 mg in 20-week run-in — N: 803 randomised of 902 — Duration: 48-week randomised phase — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: % weight change from week 20
• Trial: STEP 5 — Population: Overweight or obesity, no diabetes — N: 304 — Duration: 104 weeks — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: % weight change
• Trial: STEP 8 — Population: No diabetes — N: 338 — Duration: 68 weeks — Formulation and trial dose: SC 2.4 mg weekly vs liraglutide 3.0 mg daily — Endpoint discussed: % weight change
• Trial: STEP TEENS — Population: Adolescents 12 to under 18, obesity — N: 201 — Duration: 68 weeks — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: % change in BMI
• Trial: STEP UP — Population: Obesity, BMI ≥30, no diabetes — N: 1,407 — Duration: 72 weeks — Formulation and trial dose: SC 7.2 mg and 2.4 mg weekly — Endpoint discussed: % weight change
• Trial: SURMOUNT-5 — Population: Obesity, no diabetes — N: 751 — Duration: 72 weeks — Formulation and trial dose: SC semaglutide 1.7 or 2.4 mg vs tirzepatide 10 or 15 mg — Endpoint discussed: % weight change
• Trial: SURPASS-2 — Population: Type 2 diabetes on metformin — N: 1,879 — Duration: 40 weeks — Formulation and trial dose: SC semaglutide 1.0 mg vs tirzepatide 5, 10, 15 mg — Endpoint discussed: HbA1c change
• Trial: SELECT — Population: Age ≥45, established CVD, BMI ≥27, no diabetes — N: 17,604 — Duration: Mean 39.8 months — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: Three-point MACE
• Trial: FLOW — Population: Type 2 diabetes and chronic kidney disease — N: 3,533 — Duration: Median 3.4 years — Formulation and trial dose: SC 1.0 mg weekly — Endpoint discussed: Major kidney disease events
• Trial: SUSTAIN-6 — Population: Type 2 diabetes, 83% with CVD or CKD — N: 3,297 — Duration: 104 weeks — Formulation and trial dose: SC 0.5 and 1.0 mg weekly — Endpoint discussed: Three-point MACE
• Trial: PIONEER 6 — Population: Type 2 diabetes at high CV risk — N: 3,183 — Duration: Median 15.9 months — Formulation and trial dose: Oral 14 mg daily — Endpoint discussed: Three-point MACE
• Trial: SOUL — Population: Age ≥50, type 2 diabetes with ASCVD, CKD or both — N: 9,650 — Duration: Mean 47.5 months, median 49.5 — Formulation and trial dose: Oral up to 14 mg daily — Endpoint discussed: Three-point MACE
• Trial: ESSENCE — Population: Noncirrhotic MASH, fibrosis F2 to F3 — N: 1,197 randomised, 800 in interim — Duration: Interim at 72 weeks of a 240-week trial — Formulation and trial dose: SC 2.4 mg weekly — Endpoint discussed: Histologic response

Results and estimand adjudication

• Trial: STEP 1 — Main result: −14.9% vs −2.4% placebo — Estimand: Treatment-policy — Other estimand: Trial-product −16.9% vs −2.4% — Which figure to quote: 14.9%, naming treatment-policy and 68 weeks. This is the default semaglutide 2.4 mg figure
• Trial: STEP 2 — Main result: −9.6% vs −3.4% placebo — Estimand: Treatment-policy — Other estimand: Trial-product −10.6% vs −3.1% — Which figure to quote: 9.6% for any claim about people with type 2 diabetes. Never substitute STEP 1 here
• Trial: STEP 3 — Main result: −16.0% vs −5.7% placebo — Estimand: Treatment-policy — Other estimand: Trial-product −17.6% vs −5.0% — Which figure to quote: 16.0%, always with the note that both arms received behavioural therapy
• Trial: STEP 4 — Main result: −7.9% continuing vs +6.9% regain — Estimand: Treatment-policy — Other estimand: Trial-product −8.8% vs +6.5% — Which figure to quote: The pair together, stating change from week 20 across 48 weeks. Never quote as change from baseline
• Trial: STEP 5 — Main result: −15.2% vs −2.6% placebo — Estimand: Treatment-policy — Other estimand: Trial-product −16.7% vs −0.6% — Which figure to quote: 15.2% at 104 weeks for any durability claim. The longest evidence available
• Trial: STEP 8 — Main result: −15.8% vs −6.4% liraglutide — Estimand: Treatment-policy — Other estimand: Not retrieved — Which figure to quote: 15.8% vs 6.4%, naming liraglutide as comparator. Do not quote the placebo arm as the finding
• Trial: STEP TEENS — Main result: BMI −16.1% vs +0.6% placebo — Estimand: Treatment-policy — Other estimand: Not retrieved — Which figure to quote: 16.1% as a BMI change only. For body weight, the placebo-subtracted figure is 17.4 points
• Trial: STEP UP — Main result: −18.7% (7.2 mg), −15.6% (2.4 mg), −3.9% placebo — Estimand: Treatment-policy — Other estimand: A higher trial-product figure circulates but is unverified — Which figure to quote: 18.7% on treatment-policy. Do not publish the higher circulating figure
• Trial: SURMOUNT-5 — Main result: Semaglutide −13.7% vs tirzepatide −20.2% — Estimand: Treatment-regimen — Other estimand: Efficacy estimand reported separately — Which figure to quote: 13.7% vs 20.2%, naming treatment-regimen and the open-label design
• Trial: SURPASS-2 — Main result: Semaglutide 1.0 mg −1.86 HbA1c points — Estimand: Not named in accessible text — Other estimand: Not retrieved — Which figure to quote: The HbA1c figures only, naming semaglutide 1.0 mg. Not a weight comparison
• Trial: SELECT — Main result: HR 0.80 (0.72-0.90); 6.5% vs 8.0% — Estimand: Intention-to-treat, no formal estimand — Other estimand: Not applicable — Which figure to quote: Both together. A 20% relative reduction is 1.5 percentage points absolute over about 40 months
• Trial: FLOW — Main result: HR 0.76 (0.66-0.88); 18.7% vs 23.2% — Estimand: Intention-to-treat, no formal estimand — Other estimand: Not applicable — Which figure to quote: Both, noting the 1.0 mg dose and that the composite includes cardiovascular death
• Trial: SUSTAIN-6 — Main result: HR 0.74 (0.58-0.95), noninferiority — Estimand: Not named — Other estimand: Not applicable — Which figure to quote: Only as a safety finding. No superiority test was reported
• Trial: PIONEER 6 — Main result: HR 0.79 (0.57-1.11), noninferiority — Estimand: Not named — Other estimand: Not applicable — Which figure to quote: Only as noninferiority. The interval crosses 1.0, so no benefit was shown
• Trial: SOUL — Main result: HR 0.86 (0.77-0.96); 12.0% vs 13.8% — Estimand: Intention-to-treat, no formal estimand — Other estimand: Not applicable — Which figure to quote: The primary result only. Confirmatory secondary outcomes, including kidney, were null
• Trial: ESSENCE — Main result: Resolution 62.9% vs 34.3%; fibrosis 36.8% vs 22.4% — Estimand: Intention-to-treat, no formal estimand — Other estimand: Not applicable — Which figure to quote: Both response rates, labelled as a 72-week interim analysis. No hazard ratio exists

Evidence status and citations

• Trial: STEP 1 — Main limitation: Mostly women and White participants; short duration — Evidence status: Peer-reviewed clinical evidence — Primary citation: Wilding JPH, et al. N Engl J Med. 2021;384(11):989-1002 — PMID or DOI: PMID 33567185
• Trial: STEP 2 — Main limitation: Insulin users excluded; low cardiovascular risk — Evidence status: Peer-reviewed clinical evidence — Primary citation: Davies M, et al. Lancet. 2021;397(10278):971-984 — PMID or DOI: PMID 33667417
• Trial: STEP 3 — Main limitation: Cannot separate drug, therapy and diet effects — Evidence status: Peer-reviewed clinical evidence — Primary citation: Wadden TA, et al. JAMA. 2021;325(14):1403-1413 — PMID or DOI: PMID 33625476
• Trial: STEP 4 — Main limitation: Run-in selected for tolerance; withdrawal design carries bias — Evidence status: Peer-reviewed clinical evidence — Primary citation: Rubino D, et al. JAMA. 2021;325(14):1414-1425 — PMID or DOI: PMID 33755728
• Trial: STEP 5 — Main limitation: Small sample; limited participant diversity — Evidence status: Peer-reviewed clinical evidence — Primary citation: Garvey WT, et al. Nat Med. 2022;28(10):2083-2091 — PMID or DOI: PMID 36216945
• Trial: STEP 8 — Main limitation: Open-label; poor tolerance handled differently per drug — Evidence status: Peer-reviewed clinical evidence — Primary citation: Rubino DM, et al. JAMA. 2022;327(2):138-150 — PMID or DOI: PMID 35015037
• Trial: STEP TEENS — Main limitation: Only one participant had overweight; short follow-up — Evidence status: Peer-reviewed clinical evidence — Primary citation: Weghuber D, et al. N Engl J Med. 2022;387(24):2245-2257 — PMID or DOI: PMID 36322838
• Trial: STEP UP — Main limitation: Closed access; trial-product estimand unconfirmed — Evidence status: Peer-reviewed clinical evidence, with one unverified figure — Primary citation: Wharton S, et al. Lancet Diabetes Endocrinol. 2025;13(11):949-963 — PMID or DOI: PMID 40961952
• Trial: SURMOUNT-5 — Main limitation: Open-label; maximum tolerated doses, not matched doses — Evidence status: Peer-reviewed clinical evidence — Primary citation: Aronne LJ, et al. N Engl J Med. 2025;393(1):26-36 — PMID or DOI: PMID 40353578
• Trial: SURPASS-2 — Main limitation: Open-label; semaglutide tested at a diabetes dose — Evidence status: Peer-reviewed clinical evidence — Primary citation: Frías JP, et al. N Engl J Med. 2021;385(6):503-515 — PMID or DOI: PMID 34170647
• Trial: SELECT — Main limitation: Secondary prevention only; 27.7% women, 3.8% Black — Evidence status: Peer-reviewed clinical evidence — Primary citation: Lincoff AM, et al. N Engl J Med. 2023;389(24):2221-2232 — PMID or DOI: PMID 37952131
• Trial: FLOW — Main limitation: Few participants on SGLT2 inhibitors; mostly White — Evidence status: Peer-reviewed clinical evidence — Primary citation: Perkovic V, et al. N Engl J Med. 2024;391(2):109-121 — PMID or DOI: PMID 38785209
• Trial: SUSTAIN-6 — Main limitation: Safety trial, not powered for superiority; retinopathy signal — Evidence status: Peer-reviewed clinical evidence — Primary citation: Marso SP, et al. N Engl J Med. 2016;375(19):1834-1844 — PMID or DOI: PMID 27633186
• Trial: PIONEER 6 — Main limitation: Short median exposure; only 137 primary events — Evidence status: Peer-reviewed clinical evidence — Primary citation: Husain M, Birkenfeld AL, Donsmark M, et al. N Engl J Med. 2019;381(9):841-851 — PMID or DOI: PMID 31185157
• Trial: SOUL — Main limitation: Required cardiovascular history; kidney effects unclarified — Evidence status: Peer-reviewed clinical evidence — Primary citation: McGuire DK, et al. N Engl J Med. 2025;392(20):2001-2012 — PMID or DOI: PMID 40162642
• Trial: ESSENCE — Main limitation: Interim analysis; trial ongoing to 240 weeks — Evidence status: Peer-reviewed clinical evidence, interim — Primary citation: Sanyal AJ, et al. N Engl J Med. 2025;392(21):2089-2099 — PMID or DOI: PMID 40305708

What cardiovascular kidney and liver trials add

Weight is not the only outcome semaglutide has been tested against. Four large trials examined harder endpoints, and their results are routinely compressed into misleading single percentages.

SELECT randomised 17,604 adults aged 45 or older who had established cardiovascular disease and a BMI of at least 27, with no diabetes. Over a mean 39.8 months, the composite of cardiovascular death, nonfatal myocardial infarction or nonfatal stroke occurred in 6.5% on semaglutide and 8.0% on placebo. The reported hazard ratio was 0.80. That is the source of the widely quoted 20% reduction, and it describes a relative effect. Expressed absolutely, the difference is 1.5 percentage points over roughly three and a half years. Everyone enrolled already had cardiovascular disease, so the trial says nothing about primary prevention.

FLOW tested semaglutide 1.0 mg, not the 2.4 mg weight-management dose, in 3,533 people with type 2 diabetes and chronic kidney disease. Major kidney disease events occurred in 18.7% against 23.2%, a hazard ratio of 0.76. That composite includes kidney failure, a sustained 50% drop in kidney function, and death from kidney or cardiovascular causes. Describing it as a purely kidney outcome overstates what was measured.

SUSTAIN-6 and PIONEER 6 are frequently cited as evidence of benefit. Both were pre-approval cardiovascular safety trials designed to rule out excess risk, against a noninferiority margin of 1.8. SUSTAIN-6 reported a hazard ratio of 0.74 with a noninferiority P value and no superiority test. PIONEER 6 reported 0.79, with a confidence interval crossing 1.0. Neither demonstrates cardiovascular benefit, and neither was built to.

SOUL was designed to test superiority. It randomised 9,650 people aged 50 or older with type 2 diabetes and existing cardiovascular or kidney disease to oral semaglutide. Major adverse cardiovascular events occurred in 12.0% against 13.8%, a hazard ratio of 0.86. One detail from SOUL gets dropped constantly. The confirmatory secondary outcomes, including the kidney composite, showed no significant difference.

ESSENCE examined semaglutide in liver disease. The interim analysis covers the first 800 of 1,197 randomised participants, all with MASH and fibrosis stages F2 to F3. Steatohepatitis resolved without worsening fibrosis in 62.9% against 34.3% on placebo. Fibrosis improved without worsening steatohepatitis in 36.8% against 22.4%. These are 72-week interim results from a trial still running to 240 weeks. No hazard ratio applies, since both endpoints are binary histologic responses.

A composite endpoint is not proof of organ protection in general. Each of these trials answers one question, in one population, against one comparator. SELECT enrolled nobody without cardiovascular disease, FLOW enrolled nobody without kidney disease, and ESSENCE enrolled nobody without biopsy-confirmed liver disease. Results do not migrate between those groups, and a benefit shown in one says nothing about the others.

How semaglutide compares with tirzepatide

One randomised trial has compared these two compounds directly in adults with obesity and without diabetes.

Semaglutide versus tirzepatide. In SURMOUNT-5, 751 adults with obesity and no diabetes received the maximum tolerated dose of either drug for 72 weeks. Mean weight change was 13.7% with semaglutide and 20.2% with tirzepatide, a 6.5-point difference on the treatment-regimen estimand. The trial ran open-label, which limits how firmly the result can be read.

Three design facts shape that number. Participants received their maximum tolerated dose, meaning semaglutide at 1.7 or 2.4 mg and tirzepatide at 10 or 15 mg. The comparison is therefore not dose-matched in any simple sense. Nobody was blinded, and weight responds to expectation. Eli Lilly funded the trial, which is normal for registration studies and still worth knowing.

The 6.5-point gap is also expressed as "47% greater weight loss," which is arithmetically correct and rhetorically inflated. Dividing 6.5 by semaglutide's 13.7% produces that figure. A relative expression makes any gap sound larger, and a difference between two substantial effects needs no inflation. A relative figure should never travel without the absolute numbers beside it.

Waist circumference moved in the same direction, falling 18.4 cm with tirzepatide and 13.0 cm with semaglutide. Reaching at least 25% weight loss was achieved by 31.6% and 16.1% respectively, on the same treatment-regimen estimand as the headline figures. Mixing a treatment-regimen weight figure with a responder rate from the trial's other estimand produces an inconsistent comparison, and both sets appear in circulation.

SURPASS-2 is a different comparison and often misused. It tested tirzepatide against semaglutide 1.0 mg in 1,879 people with type 2 diabetes over 40 weeks, with HbA1c as the primary endpoint. That is a diabetes dose, well below the 2.4 mg used for weight management. The trial cannot be read as a comparison at matched maximal dosing, and its estimands could not be confirmed from the accessible text.

Our tirzepatide evidence article grades the tirzepatide side of this comparison trial by trial, including the receptor pharmacology and the trials with no semaglutide arm.

What the semaglutide safety evidence shows

Safety information for semaglutide comes from two different places, and they are not interchangeable. A label warning reflects a regulator's judgement about a possible risk. A trial adverse-event rate counts what happened in a study. Neither establishes that the drug caused the event in any individual.

Gastrointestinal effects dominate the safety picture. In the Wegovy label, 73% of injection-treated adults reported a gastrointestinal reaction against 47% on placebo. For semaglutide 2.4 mg in adults, nausea occurred in 44% against 16% on placebo. Diarrhoea affected 30% against 16%, vomiting 24% against 6%, and constipation 24% against 11%. Those rates come from the label's adverse-reaction table, whose narrative section gives vomiting as 25%. Most events were mild to moderate and clustered during dose escalation.

The boxed warning concerns thyroid C-cell tumours, including medullary thyroid carcinoma. In rodents, semaglutide caused these tumours in a dose-dependent and duration-dependent way at clinically relevant exposures. The label states plainly that whether this happens in humans is unknown. Two contraindications follow from that warning. Semaglutide is contraindicated in anyone with a personal or family history of medullary thyroid carcinoma, or with multiple endocrine neoplasia syndrome type 2. A second contraindication, often omitted, covers prior serious hypersensitivity to semaglutide or any excipient. Rodent C-cell biology differs considerably from human biology, so the warning reflects an unresolved signal and not a demonstrated human cancer risk.

Hair loss appears in the label as a trial adverse-event rate, not a warning. Two 72-week trials tested the 7.2 mg dose against both 2.4 mg and placebo. Hair loss was reported by 5.8%, 3.3% and 1.0% of those three groups. Those three figures come from one table in one set of trials. In the separate trials of 2.4 mg alone, the label reports 3% against 1%. Quoting 5.8% as a general semaglutide rate misrepresents a dose-specific finding.

Three further risks sit in the label's warnings, and the label does not treat them identically. For pancreatitis, the instruction is to discontinue if it is suspected. For gallbladder disease the wording is stronger, stating that treatment is associated with increased occurrence of gallstones and gallbladder inflammation. Rapid weight loss from any cause raises gallstone risk, so that association is not necessarily drug-specific. For acute kidney injury, the label frames the risk as a consequence of dehydration following gastrointestinal losses.

Heart rate rises by a modest amount. The label reports mean increases in resting heart rate of 1 to 4 beats per minute in adults, against placebo. A mean change describes a group and predicts nothing about an individual.

What the evidence says about stopping muscle loss and oral semaglutide

Weight returns after semaglutide treatment stops. STEP 4 established this most clearly. Participants who had lost about 10.6% during a 20-week run-in were randomised to continue or switch to placebo. Over the next 48 weeks, those continuing lost a further 7.9% while those on placebo regained 6.9%. The trial tested abrupt withdrawal under study conditions. It does not describe planned discontinuation with clinical support, which no trial in this programme examined.

Muscle loss is where confident claims outrun the evidence. Body-composition work in this field mostly measures lean mass or fat-free mass, which include water, organ tissue and connective tissue. Losing lean mass on a scan is not the same finding as losing skeletal muscle. Many studies used bioelectrical impedance in place of DXA, and the two methods are not interchangeable. In REDEFINE 1, which did include DXA, scanning covered only 7.4% of participants. Strength and physical function were not assessed. Some loss of lean mass accompanies weight reduction from any cause, which makes attributing it to a drug harder still. Evidence status: insufficient evidence for any specific claim about skeletal muscle loss on semaglutide.

Oral and injectable semaglutide are genuinely different products, and the oral category contains more than one. Rybelsus delivers 3, 7 or 14 mg daily and is approved for type 2 diabetes, not for weight management. Ozempic tablets, approved in December 2024, cover higher strengths for the same diabetes indication.

OASIS 4 tested oral semaglutide at 25 mg in 307 adults with obesity and no diabetes, a trial PubMed indexes as phase 2. The trial ran 71 weeks, with endpoints measured at week 64. Weight fell 13.6% against 2.2% on placebo under the treatment-policy estimand. The trial-product estimand gave 16.6% against 2.7%. Roughly a fifth of participants did not complete the trial. No active comparator was included, so the study supports no comparison against the injection. SOUL, covered earlier, tested oral semaglutide for cardiovascular outcomes and is a separate study with a separate purpose.

Is semaglutide FDA approved and what does compounding mean

Several semaglutide medicines are FDA approved, each for specific uses. Ozempic injection carries three separate indications. Two cover glycaemic control in type 2 diabetes, and cardiovascular event reduction in type 2 diabetes with established cardiovascular disease. A third indication arrived on 28 January 2025. It covers reducing the risk of sustained kidney function decline, end-stage kidney disease and cardiovascular death. That indication applies in type 2 diabetes with chronic kidney disease. That last one is a risk-reduction indication in diabetes with kidney disease, not an approval to treat kidney disease generally.

Wegovy injection covers cardiovascular risk reduction, chronic weight management in adults and in patients aged 12 and over, and liver disease. The MASH indication was approved on 15 August 2025 for noncirrhotic MASH with fibrosis stages F2 to F3 in adults. It came through accelerated approval, which means continued approval may depend on a confirmatory trial, currently scheduled to report in 2029. The MASH indication applies to the injection only, at 2.4 mg.

Wegovy tablets received their own approval on 22 December 2025, under a separate application. They carry the 25 mg oral strength, for weight management and cardiovascular risk reduction in adults.

Compounded semaglutide is a different category. A compounded preparation is made by a pharmacy or outsourcing facility and has not been reviewed by the FDA for safety, effectiveness or manufacturing quality. Compounding of a drug that copies an approved product is generally restricted, with a limited exception while that product is in shortage.

That shortage exception has now closed. The FDA issued a declaratory order on 21 February 2025 finding the semaglutide injection shortage resolved. Enforcement discretion then ended on two different dates: 22 April 2025 for 503A pharmacies, and 22 May 2025 for 503B outsourcing facilities. Semaglutide does not currently appear on the FDA shortage list, and it sits on no final 503A or 503B bulk drug substances list.

One pending item is widely misreported. On 30 April 2026 the FDA announced it proposes to exclude semaglutide, tirzepatide and liraglutide from the 503B bulks list. A Federal Register notice followed on 1 May 2026. The comment period closed on 30 July 2026. As of this review, no final determination has been published. A proposal is not a rule, a notice is not a regulation, and exclusion from a compounding list is not a ban on the drug. Our guide to peptide legality covers the wider compounding framework. This page explains public regulatory documents and is not legal advice.

What research grade semaglutide evidence can and cannot establish

Every clinical result above describes an approved medicine. Each was manufactured under defined controls, tested to a defined standard and reviewed by a regulator. Those trial outcomes attach to those products.

No evidence establishes that a research-grade material sold under the same compound name shares an approved product's formulation, purity, sterility, stability or biological activity. The studies that would establish it have not been run. That is a statement about what has been tested, not an accusation about any supplier.

A certificate of analysis helps, within limits. It can support identity and purity by a stated analytical method, usually HPLC, and content against a stated specification. Mass spectrometry can confirm that the molecule matches its intended structure. Safety, efficacy, sterility and correct biological function all fall outside its scope, because chemical purity and biological behaviour are separate properties. A purity figure describes chemistry and says nothing about how a material behaves in a living system, or about exact net peptide content. Our guide to reading a certificate of analysis sets out what each section means.

PrimeTime BioLabs supplies semaglutide for laboratory research only. The trial evidence on this page does not characterise that material.

What the semaglutide evidence adds up to

Semaglutide has unusually strong human evidence for weight reduction and glycaemic control, across multiple large randomised trials. SELECT supports cardiovascular risk reduction in one defined population. FLOW and ESSENCE support kidney and liver outcomes in their own narrow groups, with real limits.

Every one of those findings carries a population, a dose, a duration, an endpoint and an estimand. Strip those away and the numbers stop meaning anything, which is how one drug ends up with four different headline percentages. Weight returns once treatment comes to an end. Questions about muscle remain genuinely unresolved. The highest dose brings more weight loss and more adverse events together.