Semaglutide Rebound: What the STEP 4 Trial Actually Shows

The Study Nobody Mentions After the Before-and-After

Semaglutide works. That part isn't controversial. What's consistently omitted from influencer posts, celebrity reveals, and pharmaceutical ads is what happens when you stop. The answer comes from STEP 4 (Rubino et al., JAMA 2021) — and it isn't flattering to the "just lose the weight" crowd.

What STEP 4 Actually Found

803 adults with obesity completed a 20-week run-in phase on semaglutide 2.4 mg weekly. By the end of that lead-in, they had lost a mean of 10.6% of body weight — already clinically significant. Then researchers randomized them: continue semaglutide, or switch to placebo. Both groups received lifestyle counseling throughout.

By week 68, the divergence was stark:

• Continuers reached ~17.7% below original baseline
• Placebo-switchers retained only ~4.4% net loss
• Gap: 14.8 percentage points (P<0.001)
• Approximately 58% of run-in weight loss had returned within 48 weeks of stopping

It wasn't just the scale. Placebo-switchers diverged by +9.7 cm in waist circumference and +3.9 mmHg in systolic blood pressure versus continuers. These aren't cosmetic metrics. The SELECT trial (NEJM, 2023) established that semaglutide cuts major adverse cardiovascular events by 20% in high-risk patients — reversing those gains upon withdrawal carries measurable clinical consequences, not just aesthetic ones.

The Withdrawal Trajectory Doesn't Plateau

STEP 4's 48-week post-discontinuation window is probably optimistic. A 2025 meta-analysis by Tzang et al. in EClinicalMedicine pooled 18 RCTs (N=3,771) and found that rebound in weight, waist circumference, and blood pressure increases progressively with longer follow-up after stopping. The 58% figure from STEP 4 is a lower bound. Full regain to pre-treatment baseline appears to be the terminal trajectory — it just takes longer than any single trial has measured.

SURMOUNT-4 (JAMA, 2024) replicated the same pattern for tirzepatide: stopping reversed weight loss at similar velocity, confirming this is a class-level property of GLP-1 receptor agonists — not a semaglutide-specific quirk.

Not All GLP-1s Are Equal on the Way Down

The Tzang meta-analysis stratified by agent. Semaglutide produced significantly greater post-discontinuation rebound than liraglutide. The probable mechanism: greater peak efficacy creates more pharmacological weight suppression to lose when the drug is removed. Higher ceiling while on the drug; lower floor when you stop it.

This doesn't make liraglutide the smarter long-term choice — STEP 8 (JAMA, 2022) confirmed semaglutide's superior on-drug efficacy by a significant margin. The point is that rebound magnitude scales with how far down the drug took you. More effective drugs carry a steeper withdrawal cost.

Most People Are Already Living This Outcome

This isn't a fringe scenario. Rodriguez et al. (JAMA Network Open, 2025) documented GLP-1 RA discontinuation rates of 37–81% among US adults within one year of starting, driven primarily by cost and insurance coverage loss — not side effects, not preference, not clinical decision.

At $900–1,300/month without coverage, formulary changes, prior authorization denials, and coverage gaps push most patients off the drug before the one-year mark. The STEP 4 withdrawal trajectory isn't a theoretical edge case. It's the probable modal outcome for the majority of people who have ever started semaglutide in the United States.

The Biology Is Not Neutral on Cessation

Post-cessation rebound isn't unique to GLP-1 RAs. Any intervention producing significant fat mass reduction triggers compensatory adaptations: elevated ghrelin, suppressed leptin, increased caloric efficiency, and primed orexigenic signaling. The body defends its adipose setpoint aggressively, and it has evolved to do so far more effectively than any behavioral intervention can counter. A 2025 meta-analysis on fasting gut hormone responses confirmed similar hormonal compensation patterns following caloric restriction and exercise — making this a universal physiological challenge, not a drug-specific design flaw.

One mechanistically notable finding: a 2025 study in Obesity found that one year of structured exercise training produced greater postprandial GLP-1 secretion than continued GLP-1 RA treatment. That's interesting data. It is not a tested withdrawal-mitigation strategy, and should not be read as "just exercise instead." The populations, durations, and outcomes are not comparable in any clinically actionable way.

The Verdict

Semaglutide is a chronic medication managing a chronic disease. The evidence from STEP 4, the Tzang meta-analysis, and SURMOUNT-4 makes the maintenance requirement unambiguous. Discontinuation is not a phase transition to an independent maintained state. It's a reversal — documented, progressive, and cardiometabolically consequential.

The system pricing this drug out of reach for 37–81% of users within a year is not producing a population that kept the weight off. It's producing documented cardiometabolic deterioration at population scale. That is a policy failure with measurable cardiovascular consequences. The before-and-after photos aren't the story. The after-after is.