Was the STEP-1 Extension a randomized controlled trial?

No. It was a single-arm, observational extension of the parent STEP-1 RCT. Participants were not re-randomized at week 68. All participants (former semaglutide and former placebo groups) discontinued study drug and were followed observationally for 52 additional weeks.

How many STEP-1 participants actually enrolled in the extension?

327 of the original 1,961 participants (16.7%). Enrollment was voluntary and limited to participating sites that opted into the extension protocol, creating potential selection bias.

Did participants receive any treatment during the extension period?

No study drug was administered during weeks 68-120. Participants received standard lifestyle counseling but no pharmacotherapy. They were not prohibited from seeking other weight-management interventions outside the study.

How much weight did participants regain after stopping semaglutide?

Participants regained approximately 11.6 percentage points of body weight between weeks 68 and 120. Given the mean on-treatment loss of 14.9%, this represents roughly two-thirds of the achieved weight loss returning within one year.

Does the STEP-1 Extension prove that lifelong semaglutide is necessary?

It demonstrates that weight regain is the typical outcome after discontinuation, which supports the chronic-disease framing. However, the study was not designed to test whether behavioral interventions, dose tapering, or other strategies could prevent regain. It describes what happens, not what must happen.

Why were patients with type 2 diabetes excluded?

STEP-1 excluded T2D patients because they were studied separately in STEP-2. GLP-1 agonists have glucose-dependent effects that may alter weight dynamics differently in diabetic populations, making separate evaluation scientifically appropriate.

What statistical model was used for the primary analysis?

A mixed model for repeated measures (MMRM) was used, which handles missing data under the missing-at-random assumption without requiring explicit imputation. Approximately 90% of extension participants completed the final visit.

How does this differ from the STEP-4 withdrawal design?

STEP-4 re-randomized responders at week 20 to continued semaglutide or placebo switch, providing a controlled comparison. The STEP-1 Extension had no continuation arm and was purely observational after drug cessation, making it descriptive rather than comparative.

Did cardiometabolic improvements also reverse after stopping?

Yes. Improvements in waist circumference, blood pressure, and lipids partially reversed during the off-treatment period, though small residual benefits remained at week 120 compared to pre-treatment baseline.

Were there any predictors of who maintained weight loss?

The primary publication did not report subgroup analyses identifying predictors of successful weight maintenance after discontinuation. Individual-level trajectory data were not published, limiting clinicians' ability to counsel specific patients about their likely outcome.

Inside the STEP-1 Extension Methodology: What Most Summaries Skip

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

At a glance

| Parameter | Detail | |-----------|--------| | N (extension) | 327 of original 1,961 | | Intervention | Off-treatment observation (semaglutide discontinued at week 68) | | Comparator | Within-subject change from week 68 to week 120 | | Duration | 52-week off-treatment extension (weeks 68-120) | | Primary endpoint | Change in body weight (%) from week 68 to week 120 | | Key result | +11.6 percentage points regain after a mean -14.9% loss on-treatment | | Journal | Diabetes, Obesity and Metabolism (2022) |

Why the Extension Design Matters More Than the Parent Trial

The parent STEP-1 trial established that semaglutide 2.4 mg produced 14.9% mean weight loss over 68 weeks versus 2.4% with placebo. That finding secured FDA approval for Wegovy. The extension, however, answers a different and arguably more consequential question: what happens to the weight after the drug stops?

Most trial summaries treat the extension as an afterthought. But its design choices determine whether the "two-thirds regain" number reflects a true biological signal or a methodological artifact. Understanding those choices is essential for clinicians counseling patients about treatment duration.

Enrollment and Selection: Who Entered the Extension?

Of the original 1,961 participants randomized in STEP-1, only 327 (16.7%) enrolled in the off-treatment extension. This subset was not re-randomized. Participation was voluntary and limited to sites that opted into the extension protocol.

This creates a well-known selection problem. The 327 who continued likely differed systematically from those who did not. Possible biases cut both directions:

Participants who experienced the most dramatic weight loss may have been more motivated to continue follow-up (biasing toward greater regain if regression to the mean operates)
Participants at sites with stronger research infrastructure may have had better lifestyle support during the off-treatment period (biasing toward less regain)
Those who experienced adverse events or poor tolerability had already dropped from the parent trial and were unavailable for extension enrollment

The authors acknowledged this limitation but did not perform formal propensity-score matching between extension participants and non-participants. The primary publication reports baseline characteristics of the extension cohort that were broadly similar to the full STEP-1 population, but "broadly similar" on observed covariates does not rule out selection on unobserved factors.

The Estimand Framework: What Exactly Was Being Estimated?

Modern obesity trials increasingly adopt the ICH E9(R1) estimand framework, which forces explicit specification of what treatment effect is being targeted. The STEP-1 Extension used what can be characterized as a "treatment policy" estimand for the off-treatment period, meaning it captured weight change regardless of what participants did after stopping semaglutide.

This is a critical interpretive point. The extension did not:

Mandate that participants avoid other weight-management interventions
Restrict bariatric surgery, other medications, or structured programs
Control for changes in physical activity or caloric intake

What it measured was the net trajectory of weight after drug discontinuation under real-world-adjacent conditions (with the caveat that trial participants receiving regular weigh-ins and clinic visits already receive more structure than typical clinical practice).

The HealthRX Estimand Interpretation Matrix:

| Estimand element | STEP-1 Extension approach | Clinical implication | |-----------------|--------------------------|---------------------| | Population | Voluntary subset, not re-randomized | Generalizability uncertain | | Treatment | Discontinuation of semaglutide 2.4 mg | Tests offset, not a new intervention | | Intercurrent events | Treatment-policy (all included) | Real-world relevance high, internal validity lower | | Endpoint | % body weight change week 68→120 | Captures net rebound trajectory | | Summary measure | Mean change | Masks individual variability |

This framework matters because it distinguishes the STEP-1 Extension from a withdrawal-design RCT (like STEP-4, which re-randomized responders to continuation vs. placebo). The extension is observational in nature despite being embedded within an RCT infrastructure.

Blinding and Its Absence in the Extension

During the parent trial, participants were blinded to semaglutide vs. placebo assignment. In the extension period, blinding was effectively broken because:

All participants discontinued study drug (both semaglutide and placebo groups stopped injections)
The dramatic weight difference between groups at week 68 made assignment obvious to participants and investigators
No sham injections were administered during the off-treatment period

The absence of blinding during weeks 68-120 introduces potential behavioral confounding. Participants who knew they had been on active semaglutide (and could feel appetite returning) may have responded differently than those who knew they were on placebo. This is not a design flaw per se, as the extension was measuring a naturalistic trajectory, but it does limit causal claims about the biological inevitability of regain.

Statistical Approach: Mixed Model for Repeated Measures

The primary analysis used a mixed model for repeated measures (MMRM) to estimate mean weight change from week 68 to week 120. Key statistical features included:

No imputation of missing data: MMRM handles missingness under the missing-at-random (MAR) assumption
Observed-cases analysis: reported alongside the primary MMRM
No multiplicity adjustment: the extension had a single primary endpoint without a hierarchical testing structure

The MAR assumption is questionable here. Participants who regained weight rapidly may have been more likely to discontinue follow-up visits (missing not at random). If anything, this would bias the reported regain downward, meaning the true population-level regain could be worse than 11.6 percentage points.

The trial reported that of the 327 extension participants, approximately 90% completed the week-120 visit, which is a strong retention rate that somewhat mitigates MNAR concerns.

The Comparator Problem: No Active Control Arm

A fundamental limitation: there was no arm that continued semaglutide through week 120 in this extension. The STEP-4 trial partially addresses this by randomizing week-20 responders to continued semaglutide vs. placebo switch, showing continued weight loss with maintenance and regain with discontinuation.

But STEP-1 Extension cannot directly answer "how much regain would have been prevented by continuation?" because the comparison requires cross-trial inference. The subsequently published STEP-5 trial (104 weeks continuous treatment) provides the closest parallel, showing sustained -15.2% weight loss at two years with ongoing semaglutide.

Inclusion and Exclusion Criteria Inherited From the Parent Trial

The extension inherited STEP-1's eligibility criteria, which shaped the population in clinically relevant ways:

BMI ≥30 kg/m² or ≥27 with at least one weight-related comorbidity: standard obesity-trial criteria
Excluded: type 2 diabetes (studied separately in STEP-2), prior bariatric surgery, recent cardiovascular events
Required: at least one self-reported unsuccessful dietary effort

The exclusion of type 2 diabetes is important because GLP-1 agonists have glucose-dependent mechanisms that may alter weight-regain kinetics in diabetic vs. non-diabetic populations. The extension results should not be directly extrapolated to patients with T2D.

Cardiometabolic Parameter Reversal

Beyond weight, the extension tracked cardiometabolic markers. By week 120, improvements in waist circumference, systolic blood pressure, and lipid parameters that had accrued during treatment also partially reversed. This multi-system regression strengthens the interpretation that semaglutide-mediated improvements require ongoing drug exposure, consistent with the FDA's Wegovy prescribing information framing obesity as a chronic disease.

| Parameter | Week 0→68 (on drug) | Week 68→120 (off drug) | Net at week 120 vs. baseline | |-----------|---------------------|------------------------|------------------------------| | Body weight | -14.9% | +11.6 pp regain | Approximately -3.3% | | Waist circumference | -13.5 cm | +9.4 cm | -4.1 cm | | Systolic BP | -6.2 mmHg | +4.4 mmHg | -1.8 mmHg |

The residual benefit at week 120 (still -3.3% from baseline) suggests partial durability, but from a clinical-significance standpoint, this falls below the 5% threshold that guidelines consider clinically meaningful weight loss.

What This Trial Cannot Tell You

Whether all patients regain equally. The mean masks substantial individual variability. Some participants maintained most of their loss; others regained beyond baseline. The trial did not publish individual-level trajectory data or identify predictors of successful maintenance.
Whether structured behavioral intervention during offset could attenuate regain. The extension did not test this. The ongoing STEP-UP trial and real-world registries may eventually answer this question.
Whether dose tapering (vs. abrupt discontinuation) changes the regain curve. All participants stopped at full dose (2.4 mg weekly) without a taper protocol.
Long-term safety of indefinite continuation. The extension's value is in demonstrating the problem. Whether lifelong GLP-1 therapy is the solution requires different evidence (ongoing post-marketing surveillance, SELECT cardiovascular outcomes data).

Clinical Translation

The STEP-1 Extension's methodology, specifically its observational design, voluntary enrollment, and treatment-policy estimand, produces a result best interpreted as: "Under conditions roughly approximating clinical practice, most patients who stop semaglutide after 68 weeks will regain the majority of lost weight within one year."

This is not the same as proving that biology mandates regain. The trial was not designed to isolate biology from behavior, environment, or psychological factors. It was designed to describe what happens, and what happens is that weight returns. For prescribers, the practical implication is straightforward: discuss treatment duration expectations before initiating therapy, not after patients ask to stop.

Frequently asked questions

›

References

Wilding JPH, Batterham RL, Davies M, et al. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension. Diabetes Obes Metab. 2022;24(8):1553-1564. https://pubmed.ncbi.nlm.nih.gov/35441470/
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(14):1414-1425. https://pubmed.ncbi.nlm.nih.gov/34170647/
Garvey WT, Batterham RL, Bhatta M, et al. Two-year effects of semaglutide in adults with overweight or obesity: The STEP 5 trial. Nat Med. 2022;28(10):2083-2091. https://pubmed.ncbi.nlm.nih.gov/36356085/
FDA. Wegovy (semaglutide) prescribing information. Revised 2023. https://www.accessdata.fda.gov/drugsatfda_docs/label/2023/215256s007lbl.pdf
Jensen MD, Ryan DH, Apovian CM, et al. 2013 AHA/ACC/TOS guideline for the management of overweight and obesity in adults. Circulation. 2014;129(25 Suppl 2):S102-S138. https://pubmed.ncbi.nlm.nih.gov/24345482/