STEP-1 Subgroup Analyses: Who Responded Most and Least to Semaglutide 2.4 mg
At a glance
| Field | Detail |
|---|---|
| Trial | STEP-1 (Semaglutide Treatment Effect in People with Obesity) |
| N | 1,961 adults with BMI ≥30 (or ≥27 with ≥1 comorbidity), without diabetes |
| Intervention | Semaglutide 2.4 mg subcutaneous once weekly |
| Comparator | Matched placebo injection, both arms with lifestyle intervention |
| Duration | 68 weeks (16-week dose escalation + 52-week maintenance) |
| Primary endpoint | Percent change in body weight from baseline to week 68 |
| Key result | −14.9% semaglutide vs −2.4% placebo (estimated treatment difference −12.4 percentage points; P <0.001) |
| Registration | NCT03548935 |
Why Subgroup Data Matters More Than the Topline Number
A 14.9% mean weight loss is striking, but means hide distributions. Clinicians prescribing semaglutide 2.4 mg (Wegovy) need to know whether the drug works differently in a 62-year-old man with a BMI of 46 compared to a 34-year-old woman with a BMI of 31. The STEP-1 publication included pre-specified subgroup analyses and supplementary forest plots that address exactly this question, yet most summaries skip them entirely.
This page walks through each subgroup stratum, quantifies the differences, and flags what remains unknown.
Pre-Specified Subgroups in the Trial Design
The STEP-1 statistical analysis plan defined subgroup analyses for the primary and key secondary endpoints by:
- Sex (male vs female)
- Age (<65 vs ≥65 years)
- Race (White, Black or African American, Asian, other)
- Ethnicity (Hispanic/Latino vs not)
- Baseline BMI (<35 vs ≥35 vs ≥40 kg/m²)
- Baseline body weight (continuous, by quartile)
- Region (North America, Europe, other)
These were tested for treatment-by-subgroup interaction using ANCOVA models adjusted for baseline weight and stratification factors. The trial was not powered to detect subgroup interactions, so non-significant interaction P values do not prove equivalence. They do, however, suggest the absence of large differential effects.
Weight Loss by Subgroup: The Full Breakdown
The table below consolidates data from the STEP-1 supplementary appendix forest plots and in-text reporting. All values use the treatment-policy estimand (intent-to-treat, including data after treatment discontinuation).
By Sex
| Subgroup | Semaglutide (% change) | Placebo (% change) | Treatment difference | Interaction P |
|---|---|---|---|---|
| Female (n ≈ 1,400) | −15.8% | −2.6% | −13.2 pp | 0.08 |
| Male (n ≈ 560) | −13.2% | −1.8% | −11.4 pp | , |
Women lost roughly 2 percentage points more than men on semaglutide. This pattern appears across GLP-1 trials and may relate to sex differences in adipose distribution, GLP-1 receptor density in subcutaneous fat, or caloric intake at baseline. The interaction term did not reach significance, so sex alone should not drive prescribing decisions.
By Age
| Subgroup | Semaglutide (% change) | Placebo (% change) | Treatment difference |
|---|---|---|---|
| <65 years (n ≈ 1,700) | −15.2% | −2.5% | −12.7 pp |
| ≥65 years (n ≈ 260) | −13.8% | −1.8% | −12.0 pp |
Older adults responded nearly as well as younger participants. The absolute treatment difference was within one percentage point. This is clinically reassuring given concerns about sarcopenia in older patients on aggressive weight-loss regimens. The 2022 AGA guideline on anti-obesity medications notes that exercise co-intervention is especially important in patients over 65 to preserve lean mass, a consideration the STEP-1 lifestyle component addressed only partially.
By Baseline BMI
| Subgroup | Semaglutide (% change) | Placebo (% change) | Treatment difference |
|---|---|---|---|
| BMI 27 to <35 | −16.2% | −2.7% | −13.5 pp |
| BMI 35 to <40 | −15.0% | −2.3% | −12.7 pp |
| BMI ≥40 | −13.5% | −2.0% | −11.5 pp |
A clear gradient emerged: patients starting at a lower BMI lost a greater percentage of their body weight. In absolute kilograms, however, patients with higher baseline BMI lost more total weight (roughly 16 to 18 kg vs 13 to 14 kg in the lowest stratum). This dissociation between relative and absolute loss matters. Percentage loss drives cardiometabolic risk thresholds (the commonly cited 5%, 10%, 15% benchmarks), while absolute loss may matter more for mechanical joint loading and obstructive sleep apnea severity.
By Race and Ethnicity
| Subgroup | n (semaglutide arm) | Semaglutide (% change) | Placebo (% change) |
|---|---|---|---|
| White | ~1,050 | −15.5% | −2.5% |
| Black or African American | ~120 | −11.8% | −1.5% |
| Asian | ~65 | −14.2% | −2.1% |
| Hispanic/Latino ethnicity | ~200 | −14.8% | −2.8% |
Black participants showed a numerically smaller percentage weight loss (approximately 3.7 percentage points less than White participants on semaglutide). This finding appeared in STEP-1 and recurred in STEP-2 and STEP-3. The subgroup sizes were small, so confidence intervals overlapped substantially, and the interaction P was not significant. Still, the consistency of the signal across STEP trials warrants attention. Potential explanations include pharmacokinetic differences, baseline body composition, cultural dietary patterns affecting the appetite-suppression pathway, and differential retention. No definitive mechanistic explanation has been established.
By Region
North American participants (who made up roughly 55% of the trial) showed a modestly larger treatment effect than European participants. This likely reflects baseline BMI differences between regions (higher mean BMI in the North American cohort) and possibly different dietary contexts affecting the lifestyle intervention arm.
The Responder Distribution: Beyond the Mean
STEP-1 reported categorical responder thresholds that reveal the distribution hidden inside the mean.
| Threshold | Semaglutide | Placebo | NNT |
|---|---|---|---|
| ≥5% weight loss | 86.4% | 31.5% | 2 |
| ≥10% weight loss | 69.1% | 12.0% | 2 |
| ≥15% weight loss | 50.5% | 4.9% | 3 |
| ≥20% weight loss | 32.0% | 1.7% | 4 |
Half of semaglutide-treated patients lost 15% or more. One in three lost at least 20%. These are numbers previously associated only with bariatric surgery. The placebo arm's 31.5% reaching ≥5% loss also underscores that lifestyle intervention alone is not trivial, a point sometimes lost in the drug-versus-placebo framing.
Roughly 14% of semaglutide-treated patients lost <5% of body weight. These "non-responders" have not been well characterized in published STEP-1 analyses. Post-hoc data from the broader STEP program suggest that early weight-loss trajectory (percent lost by week 12 to 16) is the strongest predictor of final response, which aligns with the Wegovy prescribing information recommendation to reassess at 16 weeks.
Biomarker Subgroups: Glycemia, Lipids, and Inflammation
STEP-1 enrolled adults without diabetes, but baseline prediabetes status (HbA1c 5.7% to 6.4%) was a stratification factor. Among the roughly 40% of participants who met prediabetes criteria at enrollment:
- Weight loss on semaglutide was comparable to the normoglycemic subgroup (approximately −14.5% vs −15.2%).
- HbA1c improvements were larger in the prediabetic subgroup (mean reduction −0.45 vs −0.15 percentage points), as expected given more room for improvement.
- Reversion to normoglycemia occurred in 84.1% of prediabetic participants on semaglutide vs 47.8% on placebo.
Baseline lipid and CRP strata were not formally reported as subgroups in the primary publication but were explored in subsequent pooled STEP analyses. The STEP-5 long-term extension showed that cardiometabolic improvements tracked with degree of weight loss rather than baseline biomarker severity.
What the Subgroup Data Does Not Tell Us
Several gaps limit the clinical utility of these analyses.
Small cell sizes. The Black/African American subgroup had roughly 120 patients on active drug. Asian participants numbered about 65. These sizes yield wide confidence intervals and make definitive conclusions about differential efficacy impossible.
No interaction with socioeconomic status. Adherence to once-weekly injections, cost barriers, and access to the lifestyle intervention component all vary by income and insurance status. STEP-1 did not report outcomes stratified by these factors.
Lean mass composition unknown by subgroup. Total body weight loss was the endpoint. Dual-energy X-ray absorptiometry (DXA) data were collected in a subset, but subgroup-level body composition results (fat mass vs lean mass loss by age, sex, or BMI stratum) have not been published from STEP-1. This matters most for older adults and those with lower baseline BMI, where proportional lean mass loss could be clinically significant.
No genetic pharmacogenomic stratification. Unlike some oncology trials, STEP-1 did not genotype participants for GLP-1 receptor polymorphisms or other candidate loci. Early pharmacogenomic work in GLP-1 agonist response remains preliminary.
Clinical Translation: A Practical Subgroup-Based Prescribing Framework
The consistency of treatment effect across subgroups is the dominant finding, not the variation. Every pre-specified subgroup showed a treatment difference of at least 10 percentage points over placebo. This means:
- Do not withhold semaglutide based on age, sex, or BMI alone. All subgroups responded.
- Set slightly more conservative expectations for patients with BMI ≥40. They will likely lose a lower percentage but more absolute weight.
- Monitor Black patients and older adults for early trajectory. The numerically smaller effect sizes (which did not reach statistical significance) justify closer 12-to-16-week reassessment rather than a different drug choice.
- Use the 16-week checkpoint. If a patient has not lost ≥5% by week 16, the probability of meaningful long-term response drops substantially, regardless of subgroup.
Frequently asked questions
Did men or women lose more weight on semaglutide in STEP-1?
Women lost slightly more (approximately −15.8% vs −13.2%), but the treatment-by-sex interaction was not statistically significant. Both sexes showed clinically meaningful weight loss well above placebo.
Does semaglutide 2.4 mg work in patients over 65?
Yes. Participants aged 65 and older lost approximately 13.8% of body weight on semaglutide vs 1.8% on placebo. The treatment difference was within one percentage point of the younger subgroup. Lean mass preservation through exercise is an important co-intervention in this age group.
Is semaglutide less effective at higher BMIs?
Percentage weight loss was slightly lower in the BMI ≥40 subgroup (−13.5%) compared to the BMI 27 to 35 subgroup (−16.2%). In absolute kilograms, however, higher-BMI patients lost more total weight. Both groups exceeded the 10% threshold associated with major cardiometabolic benefit.
Did Black participants respond differently to semaglutide in STEP-1?
Black participants showed numerically lower percentage weight loss (approximately −11.8% vs −15.5% in White participants). The difference was not statistically significant due to small subgroup size (roughly 120 patients). This pattern appeared consistently across multiple STEP trials.
What percentage of STEP-1 patients lost at least 20% of their body weight?
32% of semaglutide-treated patients achieved ≥20% weight loss, compared to 1.7% on placebo. This level of weight loss was previously seen primarily after bariatric surgery.
Were there non-responders to semaglutide in STEP-1?
About 14% of semaglutide-treated patients lost less than 5% of body weight. Early trajectory (weight loss by week 12 to 16) appears to be the best predictor of final response. The Wegovy label recommends reassessment at 16 weeks.
Did prediabetic patients respond differently to semaglutide?
Weight loss was similar between prediabetic and normoglycemic participants. Prediabetic patients saw larger HbA1c improvements, and 84.1% reverted to normoglycemia on semaglutide vs 47.8% on placebo.
Should subgroup data from STEP-1 change which patients get prescribed semaglutide?
No. Every pre-specified subgroup showed a treatment difference of at least 10 percentage points over placebo. Subgroup data supports using the 16-week weight-loss checkpoint for all patients rather than excluding specific groups from treatment.
Was STEP-1 powered to detect subgroup differences?
No. The trial was powered for the overall treatment effect. Subgroup analyses were pre-specified but exploratory. Non-significant interaction P values suggest the absence of large differential effects, but they cannot confirm equivalence across subgroups.
How does the STEP-1 subgroup pattern compare to tirzepatide trials?
The SURMOUNT-1 trial of tirzepatide showed a similar pattern: consistent efficacy across sex, age, and BMI subgroups, with a slight BMI gradient favoring lower baseline values. Direct comparison is limited by differences in trial populations and dose ranges.
References
- Wilding JPH, Batterham RL, Calanna S, et al. Once-weekly semaglutide in adults with overweight or obesity. N Engl J Med. 2021;384(11):989-1002. PubMed
- Wegovy (semaglutide) prescribing information. Novo Nordisk. 2021. FDA Label
- Garvey WT, Batterham RL, Bhatt DL, et al. Two-year effects of semaglutide in adults with overweight or obesity: the STEP 5 trial. Nat Med. 2022;28:2083-2091. PubMed
- Jastreboff AM, Aronne LJ, Ahmad NN, et al. Tirzepatide once weekly for the treatment of obesity. N Engl J Med. 2022;387(3):205-216. PubMed
- American Gastroenterological Association. Clinical practice guideline on pharmacological interventions for adults with obesity. Gastroenterology. 2022;163(5):1198-1225. PubMed