This article was prepared with AI assistance and has not yet been reviewed by a qualified medical or scientific professional. It is for general information, not medical advice.

The most important point on a weight-loss graph may be the point where the line usually ends.

A trial reports how much weight people lost. A headline celebrates the average. Then comes a less visible question: what happens when medicine, study supervision or both are removed?

Withdrawal studies make that question difficult to ignore. They also make one thing clear: keeping weight off is not the same task as losing it. The findings should shape long-term care, not become a verdict on someone’s willpower.

First, ask what actually stopped

In the STEP 1 extension, people previously assigned semaglutide had lost an average 17.3% of their starting weight by week 68. Over the following year they regained 11.6 percentage points—roughly two-thirds of that loss. The extension analyses included 327 people. Both the medicine and the structured lifestyle support ended at week 68. STEP 1 extension.

The result is large, but its design matters. This was a follow-up, not a new trial that split people at random into different ways of stopping care. It could not isolate the loss of the drug from the loss of support.

The distinction is easy to miss. “The weight came back after the drug stopped” is an observation. “The drug’s removal explains every part of that change” is a stronger claim.

A good care plan names what will stay in place, not just what will stop.

Tirzepatide tested a different version of the question

SURMOUNT-4 kept lifestyle counseling in both study groups. After 36 weeks of tirzepatide, 670 people were randomized to continue the medicine or switch to placebo. Over the next 52 weeks, weight increased by 14.0% in the placebo-switch group and decreased by another 5.5% in the continued-treatment group. Both percentages use weight at randomization as their reference point. SURMOUNT-4 trial.

That last detail matters. A 14% increase from a lower weight is not the same as regaining 14 percentage points of original body weight. Nor is it “14% of the weight that was lost.” The starting weight used in the math matters. Similar headlines can describe different results.

Unlike STEP’s extension, this comparison supports an ongoing medicine effect while lifestyle advice remained available. It does not establish that every alternative maintenance strategy has failed. The trial also selected people who made it through the drug’s lead-in phase. They were not everyone who might start the drug.

The fair conclusion is neither “advice makes medicine unnecessary” nor “nothing besides medicine matters.” It is that continued treatment made a useful difference in this trial.

Exercise may leave something behind when supervision ends

The Copenhagen S-LiTE program explored a different kind of support. After an initial diet phase, 195 adults without diabetes were randomized to a year of exercise, liraglutide, both, or placebo. The exercise group followed a structured program, not a passing instruction to move more. Original randomized trial.

A later analysis followed people after medicine and formal exercise supervision ended. Only 109 attended the one-year follow-up. Estimated regain after liraglutide alone was 6.0 kilograms greater than after exercise alone. But regain after liraglutide alone was not clearly different from regain after combined exercise and liraglutide. Separately, the main weight and body-fat comparisons covered both the treatment year and the following year; those results did not meet the study’s prespecified correction for multiple testing. Post-treatment study.

That statistical caution belongs beside the encouraging result, not several clicks away from it. The more tests a study runs, the more chance there is of a fluke. The team had set a stricter rule to account for that. A result that misses the rule can still be a useful lead. It is not a settled answer.

There is also a useful practical distinction. Ending an exercise program’s supervision does not always end the exercise habit. Someone may keep a routine, skill or fitness that they can still use without the study team.

That is a plausible reason to investigate exercise as part of maintenance. It does not prove that these factors caused the gap, or that exercise will let a person stop the drug.

Is the body making more of its own GLP-1?

A later report from the same research program found that exercise alone increased the late phase of the GLP-1 response to a meal compared with usual activity. The article first appeared online in September 2025, despite its 2026 issue date. It is a secondary analysis of the existing program, not a new independent trial. Endogenous GLP-1 analysis.

This offers a possible piece of the biology. It does not turn exercise into a naturally dosed version of an injected medicine. A meal-related hormone response and a drug’s exposure over time are different things. The analysis also did not show that this hormone change caused better maintenance after treatment ended.

The useful next question is what exercise adds to the drug—not which one can be made to replace the other.

Three studies, three different maintenance questions

How the withdrawal studies differ

Evidence	What was removed or compared?	Main interpretive limit
STEP 1 extension	Medicine and structured lifestyle support ended	Their separate contributions cannot be isolated from this follow-up
SURMOUNT-4	Continue tirzepatide or switch to placebo; counseling continued	Does not test every possible maintenance program
S-LiTE follow-up	Observe people after different year-long treatments ended	Attrition and uncertain comparisons limit confidence

These are different experiments, not a league table of which drug leaves the least regain. The people, starting points and care in each trial differ. These three studies involved adults without diabetes. STEP, SURMOUNT, S-LiTE.

Plan the later phase while the current one is working

The practical point is simple: ask about maintenance before a treatment change becomes urgent.

What happens if access changes? What support continues if side effects require a different plan? Which measures—weight, metabolic health, strength, everyday function—will guide the next decision? How will the care team respond to early signs that the revised plan is not working?

Stopping treatment need not be the goal. The point is to plan for what comes next, whether care stays the same or has to change.

These studies also do not provide a tapering protocol. None of these studies establishes one universal way to stop medicine, nor a supplement or exercise routine that can be promised to prevent regain. Medicine changes belong with the prescribing clinician.

Readers should also know who paid for the work. The makers of the drugs funded STEP and SURMOUNT. The Copenhagen team had foundation funds and study medicine supplied to it. These ties do not make the findings false. They are part of the record. STEP 1 disclosures, SURMOUNT-4 disclosures, and Copenhagen study disclosures.

Weight loss is an outcome. Maintenance is an ongoing task. A strong treatment plan needs room for both—and should not end where the first graph does.

Related reading: Our approach to interpreting clinical evidence.

References

• Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension
• Continued Treatment With Tirzepatide for Maintenance of Weight Reduction in Adults With Obesity: The SURMOUNT-4 Randomized Clinical Trial
• Healthy Weight Loss Maintenance with Exercise, Liraglutide, or Both Combined
• Healthy weight loss maintenance with exercise, GLP-1 receptor agonist, or both combined followed by one year without treatment
• One Year of Exercise After Weight Loss Increases Postprandial GLP-1 Secretion in Contrast to Usual Activity or GLP-1 Receptor Agonist Treatment