I'll just run a 12-week semaglutide cycle, hit my goal weight, stop, and keep the result with diet.

The most common framing of GLP-1 use in cost-conscious or finite-duration adoption: a 12-week to 6-month cycle, hit the goal, stop, maintain the loss through diet and exercise. The trial data tests this scenario directly and the result is not what the framing predicts.

What STEP-4 actually showed

The Rubino 2021 JAMA trial (STEP-4) is the discontinuation extension of the STEP-1 program. Adults who had reached week 20 of semaglutide (after the 16-week titration plus 4 weeks of maintenance) were randomized to either continue semaglutide or switch to placebo through week 68. Both groups received continued lifestyle counseling throughout.

The continued-semaglutide group lost an additional ~7.9% body weight from week 20 to week 68. The placebo group regained ~6.9%. The mean weight difference between groups at week 68 was 14.8 percentage points.

The regain trajectory is the central finding: most of the regain in the placebo group happened in weeks 4-20 post-discontinuation. By week 48 off-drug the regain rate had slowed but not reversed; the placebo group never returned to pre-trial weight, but they ended at roughly two-thirds of the original loss reversed.

SURMOUNT-4 ran the same design with tirzepatide and replicated the pattern with substantially equivalent effect sizes.

Why the cycle-and-keep framing fails

The pharmacology operates on the body's defended set point, not on calories per se. Semaglutide and the broader GLP-1 / GIP-class agonists modify the threshold at which the body resists further energy storage — they shift the appetite-energy-balance equilibrium downward. When the drug is removed, the body re-defends the prior set-point. The mechanism makes this a structural property of how the drug works, not a side-effect-of-discontinuation that can be planned around.

Several specific reasons this matters:

1. Appetite restoration is not symmetrical to appetite suppression. Users on semaglutide typically describe a substantial reduction in food noise, between-meal hunger, and craving patterns. When the drug clears, the food noise and hunger return — often more intensely than baseline, because the body is also actively defending the lost weight.

2. The set-point biology is not "willpower can override." The intervention is set-point-modifying while on-drug; the intervention is absent off-drug. Asking lifestyle willpower to substitute for the missing pharmacology is asking it to do work that's qualitatively different from the work it did pre-trial. Most users who could have maintained the lower weight through diet and exercise alone wouldn't have needed semaglutide in the first place.

3. The DEXA story compounds the trajectory. Roughly 30-40% of total weight loss during GLP-1 cycles is lean mass when resistance training and protein adequacy aren't in place. The lean-mass loss is largely irreversible — even when weight regains, the regain is heavily fat-mass. So the off-drug body composition at 12 months can be worse than pre-cycle baseline even when the scale is similar.

4. The metabolic improvements partially reverse. The lipid panel improvements, A1c reductions, blood-pressure improvements, and visceral-fat reductions that often accompany GLP-1 weight loss partially reverse on discontinuation, in parallel with weight regain. The metabolic-health justification for the cycle weakens as the off-cycle period extends.

The use cases where finite-duration GLP-1 does work

Three contexts where the cycle-and-keep framing actually pays off:

• Behavioral substrate built deliberately during the on-drug period. The on-drug period is used not just for the weight loss but as the window to install genuinely durable habits — resistance training 3x/week as a habit, eating-pattern stability, sleep adequacy. The drug is the temporary support that creates time for habit-formation work that wasn't possible pre-drug. The honest framing of this is that the on-drug period was building substrate, not pharmacologically achieving the result.

• Specific time-limited goal where post-goal rebound is acceptable. A wedding, a competition, a finite physical-demand period. The goal is achievable on the cycle; the rebound after is accepted as predictable trade-off.

• Bridge to a maintenance dose. Some users transition from full-dose semaglutide (2.4 mg weekly) to a maintenance dose (0.5-1.0 mg weekly) that they remain on chronically — not full discontinuation, but reduced ongoing dosing. The maintenance evidence base is less developed but the protocol pattern is becoming more common in real-world prescribing.

In none of these cases is the "stop and the result holds" framing accurate. The framing is "stop and the maintenance is the work I now do" or "stop and I'm planning for the rebound."

The discontinuation playbook treatment

The full structured walk-through of GLP-1 discontinuation — week-by-week timeline, biomarker recovery curves, the substrate framework that determines where you actually land — is in the GLP-1 discontinuation playbook. That doc treats this question at operational depth.

Where the critic has a real point

A subset of users do hold most of their loss post-discontinuation. They tend to be users who used the on-drug period for genuine behavioral installation rather than pure pharmacological appetite suppression. The 30% or so of STEP-4 placebo-arm participants who maintained more than half their loss at week 68 demonstrate this is possible — it's just not the modal trajectory.

Some users also genuinely succeed on shorter cycles when the on-drug period is paired with intensive behavioral therapy. STEP-3 (Wadden 2021) showed slightly larger weight loss with intensive behavioral support vs basic counseling — but the gap was small enough (about 1.1 percentage points) that the IBT itself wasn't transformative. The substrate-building effect of the on-drug period is real but variable.

Where the critic loses the thread

The leap from "some users hold most of their loss" to "I will hold most of my loss on a 12-week cycle and diet" is the optimism bias the framing typically carries. The base rate from STEP-4 is that the placebo arm regained ~⅔ of the loss. Treating the modal trajectory as a personal-exception story is the most common waste of GLP-1 spend.

The right framing: if you're considering a GLP-1 cycle, plan for indefinite therapy as the modal scenario, plan the on-drug period as substrate-building for the off-drug reality, and treat "I'll just stop and keep it" as the worst case to design against — not the goal to aim for. The trial data, not the marketing, is what should anchor the planning.