← All trial deep-dives

STEP-4 deep-dive: what happens when you stop semaglutide (Rubino 2021 JAMA)

Last verified August 2026 · Phase 3a · Completed (results published April 2021) · NCT03548987 ↗

By Eli Marsden · Founding Editor
Editorially reviewed & fact-checked against primary sources · How we verify contentLast reviewed

STEP-4 (NCT03548987) was designed by Novo Nordisk to answer a single question: what happens to weight, waist, blood pressure, and physical functioning if you stop semaglutide 2.4 mg after the initial loss phase? The trial enrolled 902 adults with overweight or obesity (BMI ≥30, or ≥27 with a weight-related comorbidity) and no diabetes, then ran them through a 20-week open-label run-in of weekly subcutaneous semaglutide titrated from 0.25 mg up to the 2.4 mg maintenance dose. The 803 participants (89.0%) who reached and tolerated 2.4 mg were randomized 2:1 to continue semaglutide (n=535) or switch to matching placebo (n=268) for another 48 weeks, with monthly lifestyle counseling in both arms. Rubino and colleagues published the results in JAMA on April 13, 2021. The primary endpoint — mean percentage change in body weight from randomization (week 20) to week 68 — was −7.9% on continued semaglutide versus +6.9% on placebo, an estimated treatment difference of −14.8 percentage points (95% CI −16.0 to −13.5; P<0.001). Every confirmatory secondary endpoint moved in the same direction.

Enrollment
803
Duration
68 weeks total: 20-week open-label run-in on semaglutide 2.4 mg, then 48 weeks of randomized double-blind continuation vs placebo. In-trial observation extended to week 75.
Drug
Semaglutide 2.4 mg (Wegovy)
Population
Adults aged ≥18 with BMI ≥30, or ≥27 with at least one weight-related comorbidity (hypertension, dyslipidemia, obstructive sleep apnea, or cardiovascular disease) and no diabetes. Randomized population (n=803): mean age 46 years (SD 12), 79% female (634/803), mean baseline body weight at randomization 107.2 kg (SD 22.7), 84% white. Mean weight loss during the 20-week run-in was 10.6%. Conducted at 73 sites across 10 countries from June 2018 to March 2020.

Primary endpoint

Percentage change in body weight from week 20 (randomization) to week 68 (in-trial estimand)

Treatment arm

−7.9%

Comparator

+6.9%

Treatment difference: Estimated treatment difference −14.8 percentage points (95% CI −16.0 to −13.5; P<0.001)

Participants who continued semaglutide lost an additional 7.9% on top of run-in losses; those switched to placebo regained 6.9% during the same 48 weeks. Net swing of nearly 15 percentage points captures both continued loss vs continued maintenance and continued maintenance vs regain.

Secondary endpoints

EndpointTreatmentComparatorDifference
Change in body weight from week 20 to week 68 (kilograms, in-trial estimand)

Absolute weight trajectories diverge sharply within weeks of randomization and continue diverging through week 68.

−7.5 kg+5.7 kgNet difference of roughly −13 kg between arms over 48 weeks of maintenance
Change in waist circumference from week 20 to week 68 (cm)

Confirmatory secondary endpoint. Central adiposity tracks total body weight: maintained on drug, partially restored off it.

−6.9 cm+3.2 cmEstimated treatment difference −9.7 cm (95% CI −10.9 to −8.5; P<0.001)
Change in systolic blood pressure from week 20 to week 68 (mmHg)

Confirmatory secondary endpoint. Blood-pressure gains from the run-in were lost in the placebo-switch arm within 48 weeks.

0 mmHg+5 mmHgEstimated treatment difference −3.9 mmHg (95% CI −5.8 to −2.0; P<0.001)
Change in SF-36 physical functioning score from week 20 to week 68

Confirmatory secondary endpoint on the validated Short Form 36 Version 2 Acute. Physical functioning continued to improve on drug; declined after switch to placebo.

+1.0 point−1.2 pointsEstimated treatment difference +2.5 points (95% CI 1.6 to 3.3; P<0.001)
Cumulative ≥5% body-weight reduction from week 0 (start of run-in) to week 68

Nearly half of the placebo-switch group still ended the trial below their pre-run-in starting weight, but the rate of meeting any meaningful threshold collapsed compared with continued semaglutide.

88.7% (461/520)47.6% (119/250)Stark divergence in who held the run-in result through 48 weeks of maintenance.
Cumulative ≥10% body-weight reduction from week 0 to week 6879.0% (411/520)20.4% (51/250)Roughly 4× more participants on continued semaglutide held a clinically meaningful weight loss through trial end.
Cumulative ≥15% body-weight reduction from week 0 to week 68

The threshold most often cited as bariatric-adjacent collapsed in the placebo-switch arm.

63.7% (331/520)9.2% (23/250)Roughly 7× more participants on continued semaglutide held a ≥15% loss through trial end.
Change in HbA1c from week 20 to week 68 (percentage points)

Participants were non-diabetic at entry, so changes occur within the normoglycemic/prediabetic range, but the directional difference is consistent with the rest of the metabolic panel.

−0.2+0.1Glycemic gains from run-in preserved on continued semaglutide; small reversal on placebo.
Change in fasting plasma glucose from week 20 to week 68 (mg/dL)−1.1 mg/dL+7.6 mg/dLNet difference of about 9 mg/dL between arms.
Participants who gained weight between week 20 and week 68

More than four out of five placebo-switch participants regained weight during the 48-week maintenance phase, vs roughly one in six on continued semaglutide.

15.2% (79/520)82.4% (206/250)Direction of weight change reverses cleanly when treatment is withdrawn.

Adverse events

EventTreatment rateComparator rate
Any gastrointestinal disorder (maintenance phase only)

Rate during the 48-week post-randomization maintenance phase. Most GI events were mild-to-moderate and concentrated in participants newly continuing drug; symptoms attenuated in the placebo-switch arm as drug washed out.

49.1% (continued semaglutide)26.1% (switched to placebo)
Nausea (maintenance phase only)

Note: 422/902 (46.8%) of the full run-in cohort reported nausea during the 20-week titration, peaking during dose-escalation; rate during the maintenance phase is substantially lower.

14.0% (75/535)4.9% (13/268)
Diarrhea (maintenance phase only)14.4% (77/535)7.1% (19/268)
Vomiting (maintenance phase only)10.1% (54/535)3.0% (8/268)
Constipation (maintenance phase only)11.6% (62/535)6.0% (16/268)
Headache (maintenance phase only)7.7% (41/535)3.7% (10/268)
Cholelithiasis (serious AE, maintenance phase only)

Gallstone-related events are a known signal with rapid weight loss; absolute numbers were small in both arms.

0.9% (5/535)0.7% (2/268)
Permanent treatment discontinuation due to adverse events (maintenance phase only)

Discontinuation for AEs was essentially identical in the two maintenance arms — most tolerability issues had already resolved during the 20-week run-in before randomization.

2.4%2.2%
Any serious adverse event (maintenance phase only)

No serious AE class was concentrated in the continued-semaglutide arm. One death occurred in the continued-semaglutide arm; none in placebo (death certificate adjudicated as unrelated to drug).

5.2% (28/535)5.6% (15/268)

Clinical significance

STEP-4 is the trial regulators, payers, and clinicians cite when they describe obesity as a chronic disease requiring chronic therapy. The 14.8 percentage-point swing between continued semaglutide and the placebo switch — accumulated over 48 weeks in a population that had already lost an average 10.6% during the run-in — is not subtle. By week 68 more than four out of five participants who stopped the drug had regained weight, while only about one in six who continued did. Waist circumference, systolic blood pressure, and SF-36 physical-functioning scores all tracked the same trajectory: held on drug, partially reversed off it. The result has shaped FDA labeling, payer prior-authorization architecture (initial-approval vs continued-coverage criteria), and the clinical framing of "drug holidays" as predictably leading to regain. STEP-4 also informed the design of every subsequent maintenance trial in the GLP-1 class, including SURMOUNT-4 for tirzepatide.

Frequently Asked Questions

On average, you will regain a large fraction of what you lost within a year. In STEP-4, participants who had already lost a mean 10.6% during the 20-week run-in were randomized either to continue semaglutide 2.4 mg or to switch to placebo. Over the next 48 weeks the continued-semaglutide arm lost another 7.9%, while the placebo-switch arm regained 6.9%. That is a 14.8 percentage-point difference between staying on and stopping, accumulated over less than a year. Roughly 82% of the participants who stopped gained weight back during that window, versus 15% of those who continued. The pattern is consistent across STEP-1's extension study, SURMOUNT-4 for tirzepatide, and real-world claims-data analyses.
Not all, but a substantial fraction. At week 68 — 48 weeks after randomization to placebo — the placebo-switch group had regained 6.9% from their week-20 randomization weight. That is less than the 10.6% they had lost during run-in, so on net most participants still ended slightly below their pre-trial baseline. Whether the trajectory eventually returns to or exceeds baseline depends on individual factors and follow-up duration; the STEP-1 extension study (Wilding 2022, PMID 35441470) followed participants for one year after a longer treatment period and reported regain of roughly two-thirds of the weight that had been lost. The STEP-4 number is consistent with that trajectory.
Yes, for most patients. The trial design — abrupt switch from active drug to placebo with no taper, no behavioral intensification, no bridging medication — is essentially the worst-case version of a drug holiday, and the result was rapid, broad regain across weight, waist, blood pressure, and patient-reported physical function. Clinicians who plan an intentional pause typically do so for cost, supply, side-effect, or pregnancy reasons rather than as a maintenance strategy, and counsel patients to expect regain. There is no published evidence that intermittent dosing produces equivalent long-term outcomes to continuous therapy.
Both are withdrawal-design trials and both came out the same way directionally, but with different magnitudes. SURMOUNT-4 (Aronne 2024 JAMA, PMID 38078870) used a 36-week open-label tirzepatide lead-in followed by a 52-week randomized phase, with placebo-switch participants regaining 14.0% while continued-tirzepatide participants lost another 5.5% — a 25.3 percentage-point swing. STEP-4 used a 20-week semaglutide lead-in and a 48-week randomized phase, with a 14.8-point swing. The trials are not directly comparable because both the lead-in lengths and the maintenance durations differ, but they tell the same story: continued treatment maintains the loss; abrupt withdrawal reverses it.
STEP-4 did not test a taper, so the trial cannot answer this directly. The protocol switched participants from 2.4 mg straight to placebo. Some clinicians do step patients down through the lower doses (1.7 mg, 1.0 mg, 0.5 mg, 0.25 mg) over weeks to months, and the GLP-1 receptor pharmacology supports the idea that lower doses preserve some appetite effect at less side-effect burden. But there are no randomized trial data showing that a taper meaningfully changes the long-term regain trajectory compared with cold-stopping; the underlying biology — restoration of hunger drive and gastric emptying as drug exposure falls — applies whether the fall happens in a week or in months. If cost or supply forces a pause, ask your prescriber about a structured taper plus a tightened follow-up schedule.
Yes, in the expected direction. During the 48-week maintenance phase, 49.1% of participants on continued semaglutide reported any gastrointestinal event vs 26.1% on placebo, with nausea (14.0% vs 4.9%), diarrhea (14.4% vs 7.1%), and vomiting (10.1% vs 3.0%) leading the list. Despite the AE imbalance, permanent discontinuation rates for adverse events were essentially identical at 2.4% vs 2.2%, because the participants who could not tolerate semaglutide had already been removed during the 20-week run-in titration. Serious adverse events occurred in 5.2% of the continued-semaglutide arm vs 5.6% of placebo.

References

  1. 1.Rubino D, Abrahamsson N, Davies M, et al.; STEP 4 Investigators. 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. PMID: 33755728.
  2. 2.Novo Nordisk A/S. Study Results: Effect and Safety of Semaglutide 2.4 mg Once-weekly in Subjects With Overweight or Obesity Who Have Reached Target Dose During Run-in Period (STEP 4, NCT03548987). ClinicalTrials.gov. 2022. https://clinicaltrials.gov/study/NCT03548987?tab=results
  3. 3.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. PMID: 35441470.
  4. 4.Aronne LJ, Sattar N, Horn DB, et al.; SURMOUNT-4 Investigators. Continued Treatment With Tirzepatide for Maintenance of Weight Reduction in Adults With Obesity: The SURMOUNT-4 Randomized Clinical Trial. JAMA. 2024. PMID: 38078870.

Related trial deep-dives