The STEP program: semaglutide trial data summarized

The STEP program (Semaglutide Treatment Effect in People with obesity) was the Phase 3 clinical development program for once-weekly semaglutide in adults with overweight or obesity. The STEP 1 semaglutide trial — Wilding et al., New England Journal of Medicine 2021 — is the anchor result: across 68 weeks, the semaglutide arm recorded a mean body-weight change of −14.9% against −2.4% in the placebo arm. The most decision-relevant result, though, is arguably STEP 4, a randomized withdrawal trial measuring what happened after the compound was stopped. This page summarizes STEP 1, STEP 4 and STEP 8 as published, alongside the semaglutide compound overview.

What the STEP program was built to test

STEP was not one trial with one question, which is why quoting a single percentage from "the semaglutide trials" is a category error. STEP 1 tested the compound against placebo in adults without diabetes; STEP 2 enrolled participants with type 2 diabetes; STEP 3 added intensive behavioral therapy; STEP 4 withdrew treatment after an initial response; STEP 5 extended follow-up to 104 weeks; STEP 8 swapped placebo for an active comparator, liraglutide.

One design feature runs through all of them: both arms received a lifestyle intervention. Placebo participants got diet and physical-activity counselling and were observed inside a trial. That is why the STEP placebo arms did not sit at zero, and why measuring an active-arm figure against an untreated real-world baseline overstates the difference the trial actually found.

STEP 1: 68 weeks against placebo

STEP 1 randomized 1,961 adults 2:1 to semaglutide or placebo for 68 weeks. Eligibility required a BMI of at least 30, or at least 27 with one or more weight-related coexisting conditions; participants with diabetes were excluded. Mean baseline body weight was near 105 kg, mean BMI near 38, and roughly three-quarters of participants were female. Co-primary endpoints were percentage change in body weight to week 68 and the proportion reaching a reduction of at least 5%. As reported:

• Mean body-weight change of −14.9% versus −2.4% with placebo (−15.3 kg versus −2.6 kg); estimated treatment difference −12.4 percentage points (95% CI −13.4 to −11.5).
• A reduction of at least 5% in 86.4% of the semaglutide arm versus 31.5% of placebo.
• At least 10% in 69.1% versus 12.0%; at least 15% in 50.5% versus 4.9%.

Adverse events were dominated by gastrointestinal ones — nausea and diarrhea most commonly — typically transient and mild to moderate. Discontinuation owing to gastrointestinal events was 4.5% in the semaglutide arm versus 0.8% with placebo, and gallbladder-related disorders were reported more frequently with semaglutide. (Wilding JPH et al., NEJM 2021; PMID 33567185.)

STEP 4: the withdrawal design and the regain trajectory

STEP 4 answers a question the placebo-controlled trials structurally cannot. Conducted at 73 sites across 10 countries, it put all 902 enrolled adults through a 20-week run-in. The 803 who reached the maintenance dose were then randomized 2:1 to continue or switch to placebo for the remaining 48 weeks, through week 68, with lifestyle intervention in both groups.

Randomizing after the run-in is the methodological point: both groups had already demonstrated tolerance and response, so responder heterogeneity — the usual confounder when a treated arm is compared to a naive placebo arm — was removed before the comparison began.

The run-in produced a mean weight loss of 10.6% by week 20. From week 20 to week 68, the group that continued recorded a further −7.9%, while the group switched to placebo recorded +6.9% — an estimated treatment difference of −14.8 percentage points (95% CI −16.0 to −13.5). Waist circumference and systolic blood pressure also improved with continued treatment relative to placebo. Gastrointestinal events were reported in 49.1% of the continuation group versus 26.1% with placebo, while similar proportions in each group discontinued treatment because of adverse events (2.4% and 2.2%). (Rubino D et al., JAMA 2021; PMID 33755728.)

Read plainly: the trajectory reversed on withdrawal, and much of the run-in reduction was given back within 48 weeks. The effect was contingent on continued exposure, not a durable resetting of regulated body weight. Citing STEP 1's headline figure without STEP 4 quotes an on-treatment number as if it were permanent.

STEP 8: semaglutide versus liraglutide, head to head

STEP 8 is the program's active-comparator trial: 338 adults with overweight or obesity and without diabetes, randomized 3:1:3:1 at 19 US sites to once-weekly semaglutide, once-weekly matching placebo, once-daily liraglutide, or once-daily matching placebo, across 68 weeks. The blinding structure is often described incorrectly: the semaglutide-versus-liraglutide comparison was open-label, each active arm blinded only against its own matching placebo, and the two placebo groups pooled for analysis.

Mean body-weight change was −15.8% with semaglutide versus −6.4% with liraglutide — a difference of −9.4 percentage points (95% CI −12.0 to −6.8). Pooled placebo was −1.9%. Response thresholds separated further at the tails: at least 10% in 70.9% versus 25.6%, at least 15% in 55.6% versus 12.0%, at least 20% in 38.5% versus 6.0%.

The tolerability finding is the underreported one. Gastrointestinal adverse events were reported at closely comparable rates in both arms — 84.1% and 82.7% — yet discontinuation for any reason was 13.5% with semaglutide against 27.6% with liraglutide. Similar incidence, very different attrition: incidence and tolerability are not the same measurement. Two limitations bound it — STEP 8 was small relative to STEP 1, so its confidence intervals are wider, and an open-label comparison carries expectation effects a blinded design would not. (Rubino DM et al., JAMA 2022; PMID 35015037.)

The three trials side by side

Trial	Design	Randomized	Window	Mean body-weight change as reported
STEP 1 (NEJM 2021)	Semaglutide vs. placebo, adults without diabetes	1,961	68 weeks	−14.9% vs. −2.4%
STEP 4 (JAMA 2021)	Randomized withdrawal after a 20-week run-in	803	Week 20 → 68	−7.9% (continued) vs. +6.9% (switched to placebo)
STEP 8 (JAMA 2022)	Semaglutide vs. liraglutide, open-label active comparator	338	68 weeks	−15.8% vs. −6.4%

Later work extended the program upward: STEP UP randomized 1,407 adults with obesity 5:1:1 to a higher weekly dose, the dose used throughout the earlier trials, or placebo, over 72 weeks. Mean body-weight change was −18.7% in the higher-dose arm versus −15.6% and −3.9% — an estimated treatment difference of −3.1 percentage points against the established dose. Adverse events tracked dose upward rather than holding flat: gastrointestinal events in 70.8% of the higher-dose arm versus 61.2% and 42.8%, and dysaesthesia in 22.9% versus 6.0% and 0.5% (Wharton S et al., Lancet Diabetes Endocrinol 2025; PMID 40961952). It is a dose-response extension, not an independent replication.

Reading these numbers without over-reading them

Estimands are not interchangeable. STEP 1 reported its primary result under a treatment-policy estimand, assessing effects regardless of treatment discontinuation or rescue interventions. A trial-product estimand — modelling the effect assuming continued treatment without rescue — produces a larger figure from the same dataset. Neither is wrong; they answer different questions. Before comparing headline percentages, check the same estimand sits on both sides.

Cross-trial comparison is not a head-to-head. STEP 1's −14.9% and the tirzepatide figures from SURMOUNT-1 come from separate trials with different populations, durations and endpoints; the gap between them is not a measured difference between the compounds. The literature supports a direct comparison only where one was actually run — as in SURPASS-2, which compared the two in adults with type 2 diabetes at the semaglutide dose indicated for diabetes at the time, not the dose studied across the STEP program.

Trial supply was characterized supply. Every figure above came from a single, identity- and purity-verified clinical material. Benchwork referencing this literature only holds its comparison if the compound in hand is what the label says it is — a question of analytical documentation, not of the trial data. Merit lists semaglutide in 10 mg and 20 mg vial fills, each lot shipping with its HPLC and mass-spectrometry certificate, and any lot's certificate can be checked against the published record.

STEP 1 sized the effect against placebo, STEP 8 against an earlier GLP-1 comparator, and STEP 4 showed the effect measured in both was tied to continued exposure. This is a summary of published clinical data, not a recommendation for any research design. For research use only; not for human or veterinary use.

References

1. Wilding JPH, Batterham RL, Calanna S, et al. Once-Weekly Semaglutide in Adults with Overweight or Obesity. N Engl J Med 2021;384:989–1002. PMID: 33567185
2. Rubino D, Abrahamsson N, Davies M, et al. Effect of Continued Weekly Subcutaneous Semaglutide vs Placebo on Weight Loss Maintenance in Adults With Overweight or Obesity: The STEP 4 Randomized Clinical Trial. JAMA 2021;325:1414–1425. PMID: 33755728
3. Rubino DM, Greenway FL, Khalid U, et al. Effect of Weekly Subcutaneous Semaglutide vs Daily Liraglutide on Body Weight in Adults With Overweight or Obesity Without Diabetes: The STEP 8 Randomized Clinical Trial. JAMA 2022;327:138–150. PMID: 35015037
4. Wharton S, Freitas P, Hjelmesæth J, et al. Once-weekly semaglutide in adults with obesity (STEP UP): a randomised, controlled, phase 3b trial. Lancet Diabetes Endocrinol 2025;13:949–963. PMID: 40961952