Background
The STEP (Semaglutide Treatment Effect in People with Obesity) program evaluated subcutaneous semaglutide 2.4 mg weekly for obesity across multiple trial designs. STEP 4 was uniquely designed not to assess weight loss induction, but rather weight maintenance — specifically whether continuing semaglutide after successful initial weight loss is necessary to maintain outcomes.
This is a clinically critical question: once patients achieve target weight loss, should treatment continue indefinitely? The STEP 4 design elegantly addressed this by randomizing successful weight-losers to continue semaglutide or withdraw to placebo, revealing the drug’s role in ongoing weight regulation versus one-time loss.
Methods
Design: 68-week randomized, double-blind, placebo-controlled withdrawal trial
Phase 1 (Weeks 0–20): All 803 participants received open-label semaglutide 2.4 mg weekly (dose-escalated from 0.25 mg over 16 weeks)
Phase 2 (Weeks 20–68): Participants who achieved at least 5% weight loss during Phase 1 were randomized 2:1 to:
- Continue semaglutide 2.4 mg once weekly (n=535)
- Switch to placebo (n=268)
Population: 803 adults (BMI ≥30 kg/m²; or ≥27 with ≥1 weight-related comorbidity); no T2D
Primary endpoint: Change in body weight from Week 20 to Week 68
Key Findings
Phase 1 (Run-in Weight Loss):
- Mean weight loss during 20-week run-in on semaglutide: −10.6%
- 91.8% of participants achieved ≥5% weight loss (qualifying for randomization)
Phase 2 (Maintenance Randomization, Week 20→68):
| Outcome | Continue Semaglutide | Switch to Placebo | Difference |
|---|---|---|---|
| Weight change (W20→68) | −7.9% | +6.9% | −14.8% (p < 0.0001) |
| Total weight change (W0→68) | −17.4% | −5.0% | −12.4% (p < 0.0001) |
| ≥5% total weight loss at W68 | 88% | 48% | p < 0.0001 |
| ≥10% total weight loss at W68 | 79% | 29% | p < 0.0001 |
| Waist circumference change | Continued reduction | Regain | p < 0.0001 |
- Continuing semaglutide added an additional 7.9% weight loss after the already-achieved 10.6% run-in loss
- Placebo withdrawal led to 6.9% weight regain, erasing most of the run-in benefit
- The 14.8% divergence between groups at Week 68 represents one of the largest withdrawal-trial effect sizes for any obesity treatment
Cardiometabolic Parameters (Week 20 → 68):
| Parameter | Continue Semaglutide | Placebo |
|---|---|---|
| Systolic BP (mmHg) | −3.1 | +3.8 |
| Waist circumference (cm) | −4.0 | +3.5 |
| Triglycerides | Continued reduction | Partial reversal |
| CRP | Continued reduction | Partial reversal |
Tolerability:
- Adverse event profile consistent with STEP 1–3 (nausea most common; largely transient)
- Discontinuation rates similar between groups during the maintenance phase
Clinical Significance
- Continuous treatment required: STEP 4 definitively established that obesity-related weight gain returns after semaglutide discontinuation — obesity is a chronic disease requiring chronic pharmacotherapy, not a one-time intervention
- Weight loss compound effect: Patients continuing semaglutide achieved ~17% total weight loss from baseline, approaching bariatric surgery outcomes in a pharmacological approach
- Cardiometabolic benefit preservation: Continuing semaglutide sustained systolic BP, waist circumference, and inflammatory marker improvements — discontinuation partially reversed these gains
- Prescribing implications: The trial supports indefinite continuation of semaglutide for weight maintenance, challenging the clinical assumption that drug discontinuation post-goal is appropriate
Limitations
- Withdrawal design means placebo arm had 20 weeks of semaglutide pre-randomization — not a true naive control
- Selection bias: only those who tolerated semaglutide and achieved ≥5% weight loss during run-in were randomized; results not generalizable to non-responders
- 68-week total duration; very long-term maintenance beyond this period not established
- No T2D patients; metabolic responses in diabetic populations may differ
- All participants received dietary and exercise counseling — weight regain may differ in real-world settings without this support