STEP 12: semaglutide in Chinese adults

On 10 August 2026 The Lancet Diabetes & Endocrinology published STEP 12, a randomised, double-blind, placebo-controlled phase 3b trial of once-weekly semaglutide 2.4 mg run at 19 sites across mainland China and Taiwan. Mean bodyweight fell 12.1% on semaglutide against 2.2% on placebo at 44 weeks^[1]. The result itself is unsurprising. The reason the trial exists at all is the more interesting part.

The numbers

Of 254 people screened, 242 were randomly assigned 2:1 to semaglutide 2.4 mg (161 participants) or placebo (81 participants), each alongside a lifestyle intervention, for 44 weeks. Half the cohort (121 participants, 50.0%) was female and 47 participants (19.4%) had type 2 diabetes. Enrolment ran from 15 September 2023 to 7 May 2025^[1].

There were two coprimary endpoints. On percentage change in bodyweight the estimated treatment difference was -9.9 percentage points (95% CI -11.8 to -8.0, p<0.0001). On the proportion reaching at least a 5% reduction, 80.5% of the semaglutide group got there against 24.4% on placebo, an odds ratio of 14.8 (95% CI 7.4 to 29.6, p<0.0001). Missing week-44 data were imputed using washout multiple imputation^[1]. The trial escalated the dose every four weeks until participants reached the 2.4 mg maintenance dose, and it is registered as NCT06041217, now marked completed with registry results posted in May 2026^[2].

Why this population needed its own trial

The entry criteria answer the question. STEP 12 enrolled adults with a BMI of 24 to under 28 plus at least one weight-related complication, or a BMI of 28 to under 30 with or without complications^[1][2]. Every participant therefore sat below the WHO obesity threshold of 30 kg/m2, and most sat below the WHO overweight threshold of 25. That band is where the earlier semaglutide obesity trials did not enrol.

STEP 1, the 2021 trial that produced the 14.9% figure most people associate with the drug, required a BMI of 30 or greater, or 27 or greater with a weight-related condition, and excluded people with diabetes^[3]. STEP 6, the east Asian trial run in Japan and South Korea, required a BMI of at least 27.0 with two or more weight-related comorbidities, or at least 35.0 with one or more^[4]. Neither reached down to 24.

The reason that band matters was set out by a WHO expert consultation published in The Lancet in 2004. It reviewed evidence that Asian populations have different relationships between BMI, body-fat percentage and health risk than European populations, and found that the BMI at which risk becomes observable varies from 22 to 25 kg/m2 across different Asian populations. The consultation kept the international WHO cut-offs as the standard classification but proposed additional public-health action points at 23.0, 27.5, 32.5 and 37.5 kg/m2^[5]. Countries that define overweight and obesity at lower BMI values end up with a treatable population that global registration trials never studied. STEP 12 is the trial that fills that gap for semaglutide.

Do not line the three numbers up side by side

It is tempting to place 12.1% (STEP 12) next to 13.2% (STEP 6) and 14.9% (STEP 1) and read a gradient^[1][3][4]. That comparison does not hold. STEP 12 ran for 44 weeks; the other two ran for 68. STEP 12 enrolled a BMI band of 24 to under 30; STEP 1 started at 27 and STEP 6 at 27.0 with two comorbidities. STEP 12 included participants with type 2 diabetes, who make up 19.4% of its cohort, while STEP 1 excluded them. Baselines, durations and populations all differ, so any apparent ranking is an artefact of trial design rather than a finding about semaglutide in different bodies.

What travels better across the three trials is the placebo arm and the responder rate. Placebo groups lost 2.2%, 2.1% and 2.4% respectively, and the proportion reaching at least 5% loss was 80.5%, 83% and 86.4%^[1][3][4]. None of these trials was designed as a head-to-head, and no randomised comparison across these populations exists.

Safety, and what the abstract leaves out

Adverse events were reported in 141 of 161 participants (87.6%) on semaglutide and 61 of 81 (75.3%) on placebo, with gastrointestinal disorders the most common category. The authors describe the safety profile as consistent with what is already known about the drug^[1], which matches the pattern on our semaglutide page.

Several things are not in the published abstract. Individual event rates, treatment discontinuations and subgroup results are not broken out. Body composition is largely absent: the registry lists waist circumference, BMI, waist-height ratio, bodyweight in kilograms and blood pressure among the secondary endpoints, but no imaging-based measure of fat distribution^[2]. STEP 6 did run computed tomography in a subset and reported a 40.0% reduction in abdominal visceral fat area on semaglutide 2.4 mg against 6.9% on placebo^[4]. Given that the case for lower BMI thresholds rests on body-composition differences, that is the gap worth watching in the full paper.

Two more caveats. The trial was funded by Novo Nordisk A/S, and several co-authors are Novo Nordisk employees and shareholders, which the paper discloses^[1]. And 44 weeks is a short horizon. It says nothing about what happens after treatment stops, a question the earlier semaglutide literature answers uncomfortably.

Practical context

STEP 12 does not change how semaglutide is classified anywhere we cover. It remains a prescription-only medicine across the EU, EEA, UK and US, and the per-country detail sits on the semaglutide regulation pages. This is a registration-supporting trial for a market whose own clinical guidelines draw the overweight and obesity lines lower than the WHO does.

The wider point applies beyond China. Where a country defines obesity matters as much as what a drug does, because it decides who is eligible to be treated and which trials a regulator will accept. Readers who want the underlying pharmacology and the full trial history can start on the semaglutide overview and the regulation pages. This article is educational reporting on published trial data, not medical advice, and decisions about any of these medicines belong with a clinician who knows your history.